Migration and stratification

Abstract

Migration and stratification are increasingly intertwined. One day soon it will be impossible to understand one without the other. Both focus on life chances. Stratification is about differential life chances - who gets what and why - and migration is about improving life chances - getting more of the good things of life. To examine the interconnections of migration and stratification, we address a mix of old and new questions, carrying out analyses newly enabled by a unique new data set on recent legal immigrants to the United States (the New Immigrant Survey). We look at immigrant processing and lost documents, depression due to the visa process, presentation of self, the race-ethnic composition of an immigrant cohort (made possible by the data for the first time since 1961), black immigration from Africa and the Americas, skin-color diversity among couples formed by U.S. citizen sponsors and immigrant spouses, and English fluency among children age 8–12 and their immigrant parents. We find, inter alia, that children of previously illegal parents are especially more likely to be fluent in English, that native-born U.S. citizen women tend to marry darker, that immigrant applicants who go through the visa process while already in the United States are more likely to have their documents lost and to suffer visa depression, and that immigration, by introducing accomplished black immigrants from Africa (notably via the visa lottery), threatens to overturn racial and skin color associations with skill. Our analyses show the mutual embeddedness of migration and stratification in the unfolding of the immigrants' and their children's life chances and the impacts on the stratification structure of the United States.

1. INTRODUCTION

Migration and stratification are intimately and irrevocably linked, sharing a core focus on what Weber (1922) insightfully called life chances. Stratification is about differential life chances - who gets what and why - and migration is about improving life chances - getting more of the good things of life. Moreover, the long reach of stratification is visible in migration, and, concomitantly, migration effects are visible in the stratification structures at both origin and destination.

All the grand themes in the study of social stratification find expression in the migration process – discrimination, the pervasive effects of race and gender, the struggle of body and soul to survive, and the march to equality and full membership in society. And all the grand themes in the study of migration involve, in one way or another, stratification mechanisms – who is allowed to migrate, who actually migrates, how they fare in the destination society, what happens to their children, what happens to those left behind and to natives of the new country.

Since Weber's (1892) pioneering examination of Polish workers in Germany and, later, Thomas and Znaniecki's (1927) pathbreaking work, The Polish Peasant in Europe and America, sociologists and other social scientists have explored crucial aspects of the migration-stratification link, providing innovative ideas and theories – for example, self-selection, cumulative causation, contexts of reception, modes of incorporation, segmented assimilation, ethnic enclave, options for ethnic identity, tied movers, oppositional culture – and reinvigorating scholarship on ascriptive factors in migration – gender and race, for example – and differential life chances among migrants – in health and healthcare, housing, employment, and earnings.¹

This paper examines stratification processes in six dimensions of immigration to the United States which exemplify the links between migration and stratification: (1) U.S. government processing of new immigrants, in particular the phenomenon of lost documents, which can wreak havoc on carefully made plans and lengthen the visa process; (2) depression due to the process of applying for an immigrant visa; (3) presentation of self among new immigrants; (4) racial composition of new immigrants; (5) skin color and spouse selection among U.S. citizen sponsors of immigrant spouses; and (6) English fluency among adult new immigrants and their young children. Each of the six dimensions is evocative of the grand questions of migration and stratification, and these analyses represent an early step on the path to further exploration of their interconnections.

The six analyses reported in this paper also represent early work in a new generation of research on questions or aspects of questions for which there was a lively oral tradition but until recently little data, including aspects of classic questions such as those embodied in the last three of the six analyses as well as relatively new questions such as the first three. Indeed, questions such as the ones examined in this paper were raised again and again in immigration panels and workshops dating to the late 1970s, leading the panels to propose and progressively sharpen a new design for a large-scale data collection project on immigration, the New Immigrant Survey – which now provides the data to address those questions. In the spirit of Abbott (2004) and using his words, “new ideas” inspired “new data”, and the two together are inspiring new research, of which the present paper is an example.

The New Immigrant Survey (NIS) is a set of planned longitudinal studies of several cohorts of U.S. legal immigrants. To date, the NIS has carried out a short pilot panel study of the Fiscal Year 1996 cohort and two surveys of the Fiscal Year 2003 cohort.

It is not difficult to imagine a wealth of new research undertaken not only by immigration researchers but also by scholars across diverse fields who find in the new data the possibility of addressing longstanding questions, even foundational questions, much as is occurring with internet blogs and networking sites. Moreover, because the data are massively rich, one can envision dozens of articles, dissertations, and books on each of many topics. Finally, because the data are longitudinal, it will be possible to observe intertemporal dynamics. To illustrate, the third analysis in this paper – on presentation of self – is inspired by Goffman's (1959) foundational insights and the quintessential American possibility of self re-invention, and represents a simple first step into what could become a key empirical ingredient of several research careers.

The data used in this paper are drawn from the baseline survey of the immigrant cohort of 2003 (NIS-2003-1). The data enable examination of race, gender, religion, origin country, and language in the behavior and activities of four sets of actors: U.S. government personnel, U.S. citizen sponsors of immigrants, adult immigrants, and the children of immigrants, including those born in the United States.

Two themes permeate the paper. First is the life chances of individual immigrants. Second is the effect of immigration on the stratification structure of the United States. Together these encapsulate substantial segments of both the migrant experience in the United States and the social effects of immigration. Put differently, the stratification structure grows out of the actions and experiences of many people, and so do migration flows and the incorporation of immigrants.²

2. THEORETICAL AND EMPIRICAL FRAMEWORK

The work reported in this paper can be situated within the general social scientific theory of migration, a theory which addresses the selection, adaptation, and impacts of immigrants, together with the adaptation and impacts of their children, and which has developed in a series of contributions by diverse scholars (see, for example, Alba and Nee 2003; Bean and Stevens 2003; Jasso and Rosenzweig 1990; Massey et al. 1993; Portes and Rumbaut 2006). In the subsections that follow, we examine the basic actors and elements in the U.S. immigration process, and then consider their relevance to stratification processes and the ways that they combine to define substantively appropriate samples and subsamples or appear as explanatory factors in the six analyses. We take special care describing the main elements in the immigration process, as these shape the environment faced by prospective immigrants and new immigrants and thus will play important parts in the analyses to follow. We also describe major features of the data.

2.1. Basic Actors in the U.S. Immigration Process

Four kinds of actors play parts in the U.S. immigration process. The first is the U.S. government, which processes all legal immigrants, via the personnel who staff the agencies responsible for immigrant visa processing – the Department of State (DOS), the Department of Homeland Security (DHS), and the Department of Labor (DOL). The second is the U.S. resident or firm that sponsors the largest set of immigrants, namely, those with family or employment visas -- who together constitute almost 80 percent of the approximately one million new legal immigrants every year. The third is, of course, the immigrant him- or herself. The fourth is the young children of immigrants, including children born in the United States.

In general, everyone who comes in contact with a prospective immigrant – e.g., everyone involved in visa processing at a variety of U.S. government agencies in the United States and around the world -- affects the new immigrant's life chances. Thus, the first two actors shape the life chances of the last two, and, with them, also shape the life chances of many others, including, notably, those left behind in the origin country and the natives of the destination country.

2.2. Basic Elements of the U.S. Immigration Process

A growing insight in immigration scholarship is that immigrant behavior cannot be understood without understanding immigrants' legal status in the United States -- how they came and whether they have the coveted “green card” and, if so, how they got it, in the face of numerous obstacles and the daily deportations reported in the press (Smith and Edmonston 1997). For example, understanding labor force attachment and work ethic requires information about work authorization; understanding home ownership requires understanding the risk of deportation; and understanding the children of immigrants requires understanding whether they have a claim to U.S. citizenship. Similarly, understanding the “emotional costs” of migration (Levine, Hill, and Warren 1985:3) requires understanding the process by which immigrants reach the United States and acquire their immigration status. This tight link between the immigrant biography and migration outcomes is exemplified by the nine stories which open Portes and Rumbaut's (2006) examination of American immigration. As well, it is increasingly appreciated that a move from illegal to legal represents a highly consequential upward social mobility.

Immigrant Class of Admission

U.S. immigration law provides procedures by which persons from other countries may apply for and obtain lawful permanent residence (LPR) in the United States, a legal status which authorizes foreign-born persons to reside permanently in the United States, to engage in any occupation except those reserved for citizens, and, after satisfying residence and other conditions, to become citizens of the United States (and then engage in any occupation except President and Vice-President of the United States, which are reserved for native-born citizens, as established by the U.S. Constitution, in Article 2, Section 1, Clause 5, and the Twelfth Amendment).³

LPR classes of admission are of two main types, numerically unlimited and numerically limited. Numerically unlimited LPR is granted to the spouses, minor children (under age 21), and parents of adult U.S. citizens (a set collectively called “immediate relatives of U.S. citizens”). Numerically limited LPR is granted to three main categories of immigrants: (1) family immigrants, comprised of the adult children and siblings of U.S. citizens (a set collectively called “close relatives of U.S. citizens” to distinguish them from “immediate relatives of U.S. citizens”) and the spouses and unmarried children of LPRs; (2) employment immigrants, comprised of five subcategories; and (3) diversity immigrants (winners of the lottery visas designated for persons from countries underrepresented in recent immigration). Two additional categories of LPR admission have subsets of both numerically limited and numerically unlimited type. These are (4) humanitarian immigrants, including refugees, asylees, and parolees (RAP) and (5) legalization immigrants, that is, illegal immigrants who are becoming legal, including registry-provision immigrants who qualify in virtue of length of residence) and cancellation-of-removal immigrants, plus immigrants targeted by special legalization legislation (such as the Nicaraguan Adjustment and Central American Relief Act of 1997, or NACARA).⁴

The number of numerically limited LPR visas granted annually is about 226,000 to family immigrants, 140,000 to employment immigrants, and 50,000 to diversity immigrants. The family and employment visas are also subject to a country ceiling of 7 percent of the total. The exact number of numerically limited family and employment visas available each year is published in the Visa Bulletin issued monthly by the State Department and in the Annual Flow Report – U.S. Legal Permanent Residents and the Yearbook of Immigration Statistics issued by the Department of Homeland Security (in the latter publication only in the years before the narrative text was eliminated in 2004, such as the 2003 Yearbook of Immigration Statistics, Appendix 1). The total number of persons granted LPR is currently about a million a year.⁵

Visa Sponsorship – Sponsored and Nonsponsored Immigrants

Most family and employment immigrants require a sponsor. In the case of family immigrants, the sponsor is the relative who is already a citizen or legal permanent resident of the United States (for example, the U.S. citizen spouse or parent of a prospective immigrant). In the case of employment immigrants, the sponsor is the employing individual or firm. The sponsor files the initial petition that establishes the prospective immigrant's eligibility and starts the visa process. The requirement for a sponsor may be waived in certain cases. In the family visa classes, the sponsor requirement may be waived for the widow(er) and child of a deceased U.S. citizen (in a marriage that had existed for at least two years before the U.S. citizen's death) or for the spouse and child of an abusive citizen or LPR. In the case of employment visas, the sponsor requirement may be waived for certain classes of immigrants, including investors as well as immigrants of great renown. Sponsored immigrant cases thus require both the sponsor's petition and the prospective immigrant's application; nonsponsored immigrant cases in general require only the prospective immigrant's application.⁶

Principals and Accompanying Relatives

The “principal” is the person who qualifies for the visa. The three categories of immediate relatives of U.S. citizens – spouse, parent, minor child – are for principals only.⁷ All other categories provide LPR visas not only for the principal but also for the spouse and minor children of the principal, except for the category for spouses of LPRs and a few categories designated for “unmarried” principals, in which case “accompanying relative” visas are available only for minor children. Accompanying relatives are also called “derivatives”. The characteristics of the principal are the key characteristics in an immigration case (although, as will be seen, sponsor characteristics may also play a part), and thus in our examination of lost documents, below, in which the case is the unit of analysis, the sample is restricted to principals. Of course, if documents are lost, this affects everyone in the case, principal as well as nonprincipals.

New Arrivals and Adjustees

Prospective immigrants may apply for legal permanent residence abroad or in the United States, if they are already residing in the U.S., having been admitted with a temporary nonimmigrant visa or possibly having gained entry surreptitiously (“entry without inspection” or “EWI”). If abroad, they apply at an embassy or consular post of the State Department; if in the United States, they apply with U.S. Citizenship and Immigration Services (USCIS), an agency of the Department of Homeland Security (and, prior to March 2003, with the Immigration and Naturalization Service (INS), an agency of the Department of Justice). Applicants for employment-based LPR who are already living in the United States may elect consular processing. Applicants for lottery visas start the application process on the internet, and if selected as a possible winner continue the application process where they are residing, either abroad or in the United States.⁸

Before 1977, only persons from the Eastern Hemisphere could apply to adjust status to LPR in the United States, and not surprisingly the proportion adjusting status has increased steadily since then. During the ten-year period from 1996 to 2005, the number of adjustees exceeded that of new arrivals in every year except three (1998, 1999, and 2003, years in which administrative and processing conditions produced large backlogs in immigrant visa processing in INS offices), and for the entire period, the proportion adjustees was 55.8 percent, increasing to 64.7%, 59%, 57.9%, and 59.1% in 2006 to 2009, respectively.⁹

The set of new arrivals includes two subsets who are already living in the United States, the employment immigrants mentioned above who choose consular processing and some illegal immigrants who do not qualify for adjustment of status and go through the visa process as if they are living abroad.

Venue of Immigrant Visa Processing

Responsibilities for immigrant visa processing are shared by the Department of Homeland Security and the Department of State, with employment-based applications also involving the Department of Labor. As discussed above, in sponsored immigrant cases, the sponsor files the initiating petition with USCIS, and the prospective immigrant files either with State or with USCIS, depending on place of residence, eligibility to adjust, and employment immigrants' venue choice. Initial diversity applications are processed for the lottery at the Kentucky Consular Center operated by the State Department. Cases involving a numerically limited visa in which there is a backlog are sent to the National Visa Center operated by the State Department in Portsmouth, New Hampshire.

These processing practices imply that, in general, visa processing venue differs (1) between new arrivals and adjustees, (2) between sponsored and nonsponsored immigrants, (3) between numerically limited and numerically unlimited visas, and (4) between diversity and non-diversity visas. Thus, the combination of visa type and information on arrival/adjustment is useful for discerning processing venue.¹⁰

Accordingly, immigrant cases may straddle two venues. For example, family and employment cases may involve a sponsor submitting documents in the United States and the prospective immigrant submitting documents abroad. And it is possible to trace the case file's journey across State and USCIS facilities. To illustrate, in the case of a sponsored family immigrant, the sponsor's petition goes to USCIS and upon approval, if the visa is numerically limited and there is a backlog, the petition goes to the National Visa Center; when the visa becomes available, the sponsor's petition goes to State if this is a new-arrival case or to USCIS if it is an adjustee case. Meanwhile, the prospective immigrant's application is filed with State if it is a new-arrival case or wtih USCIS if it is an adjustee case.

The foregoing information can be distilled into two broad-brush generalizations:

• §New-arrival immigrants are always processed by State, with USCIS doing some processing for sponsored immigrants.
• §Adjustee immigrants are always processed by USCIS, with State-NVC doing some processing for numerically limited visas in which there is a backlog and State-KCC doing some processing for diversity immigrants.

Duration of Visa Processing

The processing period, from filing of the first document to admission to legal permanent residence, has two components. The first component applies only to numerically limited visas, and it refers to the time waiting in the queue for an immigrant slot (visa number) to become available for the particular combination of visa category and origin country. This component (the visa number wait) starts on the date of the filing of the first document in the case (called the “priority date”) and ranges from zero time (for visa-country combinations which are “current”) to twenty years or more (for visa-country combinations for which there are large backlogs). The case of persons immigrating from the Philippines as siblings of U.S. citizens illustrates the upper extreme (in April 2011 visas became available for applicants with a priority date of 8 March 1988 – a waiting period of 23 years); at the other extreme, visas in some of the employment-based categories (such as that for priority workers, including world-renowned scientists) are available immediately (DOS, Visa Bulletin).

The second component consists of the processing that all cases undergo, including document checks, background checks, and personal interviews. Duration of this component varies across USCIS offices and consular posts, depending on caseload and staffing. For example, the processing time at USCIS offices can increase if personnel are deployed to other duties, such as processing naturalization cases. Finally, lottery immigrants receive somewhat expedited processing, as each year's winning visas have to be used within a single fiscal year; for example, winners in the DV-2011 lottery (who applied in the fall of 2009 – and were among the 12 million principals who submitted applications covering 16.5 million prospective immigrants -- and were notified in the summer of 2010 that they had won) must complete all processing by 30 September 2011 (the end of Fiscal Year 2011).

Previous Illegal Experience

New LPRs may have spent time in the U.S. illegally, either immediately before acquiring LPR or at some earlier time. A foreign-born person becomes illegal in one of three main ways: (1) surreptitious entry; (2) overstaying a temporary visa; and (3) working without authorization. Previous illegal experience may be discerned from the immigrant visa and the nonimmigrant visa. As discussed above, some immigrant visas are explicitly legalization visas (including registry, cancellation-of-removal, and NACARA visas). Regardless of the type of immigrant visa – a legalization visa or, say, a family or employment visa – new immigrants who adjust from the EWI form of illegality are given a special code in the adjustee nonimmigrant visa field (EWI or WI). In recent years, a new code has appeared, a code for unknown (UU, sometimes UN), as well as a tendency to leave the field blank.¹¹ It is widely believed that both the UU code and missing data are a euphemism for illegal status. Finally, if the nonimmigrant visa is a visitor for pleasure visa (B2) and the most recent recorded entry is six years prior to LPR, it is reasonable to believe the person has overstayed the visa (Warren 2003, unpubl).

Accordingly, there are five types of adjustees, those who had a valid temporary visa, those who entered without inspection, those for whom a code of unknown is entered, those for whom there is no code, and those who overstay a tourist visa. Of course, persons adjusting from a valid temporary visa may have had a stint of illegal experience in the past. New arrivals may have been illegal immediately prior to obtaining LPR or at some previous time.

Pathway from a Legal Temporary Visa to Legal Permanent Residence

Most legal temporary residents have no claim on a future immigrant visa. But some are virtually guaranteed LPR and others have a smoother pathway. Persons who are admitted with refugee status or granted asylee status or who enter with a nonimmigrant K visa (for fiancé(e)s or for spouses whose application is pending) are virtually assured of progressing to LPR, provided, of course, that no impediments arise. Persons with a select type of temporary visa, such as the H-1B and L-1 visas for specialty workers and intracompany transferees, respectively, while not assured of a future LPR visa, nonetheless have the advantage that they are not required to prove that they have no intention of abandoning their country of residence in order to qualify for a temporary visa.

Conditional Legal Permanent Residence

Two sets of immigrants receive conditional visas at LPR. These are (1) spouses of U.S. citizens and of LPRs in marriages of less than two years' duration, and (2) employment-based investor immigrants. The visas are conditional for two years and a special application is made for removal of the conditionality restrictions.

2.3. Immigrant Visa Characteristics and Their Stratification Relevance

Combining Visa Characteristics

We begin by combining the visa characteristics highlighted in the previous section. As shown in Table 1, not all levels of all characteristics can occur together. For example, as noted above, not all visa types require a visa sponsor, and not all visa types permit an accompanying spouse. Similarly, immigrants with legalization visas unambiguously had previous illegal experience, although immigrants with all other kinds of visas may also have had previous illegal experience.

Table 1.

Links Among Immigrant Visa Characteristics

Immigrant Class of Admission	Numerically Limited	Requires Sponsor	Provides Visa for Spouse	New Arrival or Adjustee	Previous Illegal Experience
Immediate relative of U.S. citizen	No	Almost all	No	Both	Possible
Family preference	Yes	Almost all	Some	Both	Possible
Employment preference	Yes	Almost all	Yes	Both	Possible
Diversity	Yes	No	Yes	Both	Possible
Humanitarian	Some	No	Some	Mostly Adjustee	Possible
Legalization	Some	No	Some	Mostly Adjustee	Yes

Notes: In the process for obtaining LPR, the term “sponsor” is used in two senses, to designate both (1) the person who provides eligibility for a family-based or employment-based visa and “petitions for” the immigrant, submitting a special form (such as I-130 or I-140); and (2) a person who signs an affidavit of support (I-864) for the immigration applicant. In this paper, the term “sponsor” pertains exclusively to the first kind of sponsor, who may be called the visa sponsor. For further details, see the Glossaries on the DHS and DOS websites. Given the intricacies of U.S. immigration law, there is considerable variation, as indicated in the table. For example, some humanitarian and legalization LPR visas have been available to new arrivals, notably for Amerasians related to U.S. soldiers and for dependents of IRCA legalization immigrants. The NIS-2003 sample does not include any respondents with those visas; thus, in the NIS, all immigrants with humanitarian and legalization visas are adjustees.

Generating Subsamples and Explanatory Factors

The characteristics described in the previous section and listed in Table 1 are used to construct substantively appropriate subsamples for some of the analyses; in others they operate as explanatory factors. To fix ideas, consider two examples.

First, consider the distinction between principal and accompanying relative. In the analysis of lost documents, the case is the unit of analysis and thus the sample is restricted to principals. In the analysis of declaring oneself the principal, it is substantively appropriate to conduct separate analyses for principals and nonprincipals. In the other analyses, the visa categories used as explanatory factors distinguish between principals and nonprincipals.

Second, consider processing venue. The discussion above indicates that there are three subsets of processing venue, relevant to the prospective immigrant's life chances and in particular the possibility of lost documents: (1) processing by the State Department only; (2) processing by INS/CIS only; and (3) processing by both. Table 2 provides the processing venue circa 2003 by visa type and whether the immigrant is a new arrival or an adjustee. In 2003 employment-preference adjustees were processed by INS/USCIS alone because visas were not backlogged (DOS, Visa Bulletin, various issues).

Table 2.

Processing Venue for New Immigrants in 2003

Immigrant Class of Admission	State Only	INS/CIS Only	Both State & US/CIS
Immediate relative of U.S. citizen
New arrival			X
Adjustee		X
Family preference			X
Employment preference
New arrival			X
Adjustee		X
Diversity
New arrival	X
Adjustee			X
Humanitarian		X
Legalization		X

Notes: The table depicts venue of LPR visa processing. However, all new arrivals are inspected at the U.S. port of entry by an official of the Customs and Border Protection unit of DHS. Also, see note on humanitarian and legalization immigrants in Table 1.

Visa Type and the Human and Social Capital of Immigrants

The human and social capital of the new immigrant and the immigrant's children, as well as their prospects for integration into the United States, may be closely linked to the type of visa. Family immigrants already have a foothold in the United States -- and a counselor and advocate as well as a fountain of information on job search, housing search, medical care, etc. Obviously, the extent of this foothold and the social capital it signals varies with the type of relationship (blood kin or marital kin) and its closeness (e.g., spouse versus sibling) and whether the U.S. kin is a citizen or not. In particular, immigrant spouses of U.S. citizens may both be intensely screened and also acquire a readymade American network (Jasso and Rosenzweig 1995); additionally, these mechanisms may be intensified among spouses of native-born U.S. citizens, who thus may experience “quicker social integration” (Bean and Stevens 2003:176). Employment immigrants already have a job and an employer. Humanitarian immigrants may receive various kinds of pecuniary and nonpecuniary assistance from specialized resettlement agencies. In contrast, lottery immigrants may have nothing except their own resources – which, however, may not be inconsiderable, given the schooling and occupational requirements for eligibility.¹²

Visa Type and the Citizenship Stratification Structure

Immigrants who acquire LPR as spouses of U.S. citizens are entering a household that already has an adult U.S. citizen. They will be in a quite different situation from immigrants entering households without a single adult U.S. citizen. Other immigrants acquiring LPR as relatives of U.S. citizens will also have close access to a U.S. citizen, whether or not they reside in the same household.

Visa Type and Pioneer Immigrants

Visa type signals a further stratification-relevant distinction – between pioneer immigrants (including marital, employment, humanitarian, and diversity immigrants) and subsequent family immigrants (such as parents, siblings, and children of previous immigrants). Pioneer immigrants are thought to be more intensely positively self-selected than consanguineous family immigrants, although these distinctions may sometimes blur, as when the employment sponsor is actually a relative.¹³

Visa Type and Social Mobility

All new LPRs are moving up in the citizenship stratification structure. A move from illegality to legality is a move up, as is a move from a temporary to a permanent visa or a move from no visa (as with true new arrivals) to LPR (see Bean and Stevens 2003:111–112; Jasso, Massey, Rosenzweig, and Smith 2008; Jasso and Rosenzweig 1990; Massey et al. 1987; Powers and Seltzer 1998; Powers, Kraly, and Seltzer 2004). Some LPRs are making large jumps – notably those who are making the transition from illegality to LPR, bypassing a legal temporary visa. Others have previously made the transition from illegality to legal nonimmigrant (including, for example, some asylees) and are now moving from legal nonimmigrant to LPR.

New Arrivals, Adjustees, and Two Sources of Stress

LPR applicants and new legal permanent residents are subject to two distinct sources of stress, (1) stress associated with acquiring LPR (visa stress), and (2) stress associated with adjusting to a new country (migration stress). In general, new arrivals endure the two sources of stress sequentially, first going through visa stress (in their home country) and then going through migration stress; adjustees, on the other hand, go through both kinds of stress simultaneously (Jasso, Massey, Rosenzweig, and Smith 2005). Thus, in a conjecture reminiscent of the classic insight of Simmons and Blyth (1987) about the stress adolescents face if they must go through puberty and a school transition simultaneously, we would expect adjustees to have a more difficult time than new arrivals in the period immediately preceding LPR. Of course, there are many special cases that operate differently – for example, refugees and asylees may endure very little (LPR) visa stress because for them the daunting part of the migration process was obtaining the initial refugee or asylee status, with the subsequent adjustment to LPR being somewhat pro forma. Similarly, an LPR visa applicant who is a long-time “temporary” resident (say, someone who spent ten years on a student visa and is now in the sixth year of a temporary work visa) may have completed the process of adjustment to the United States before beginning the LPR visa application process.¹⁴

Remark on Two Meanings of “Status”

In the foregoing discussion, the word “status” has been used in the meaning of “condition” – for example, legal status, refugee status, asylee status, adjustment of status, immigrant status (joining other similar uses in sociology, such as “marital status”). The word “status” is also used in the “condition” sense to refer to a person's legality; for example, compliance with the terms of a visa “maintains status” and the visa holder remains “in status”, while violating the terms of a visa renders the visa holder “out of status.” This being a paper that aspires to link migration and stratification, the word “status” will also be used in the sense of “prestige” or “honor” or any of the synonyms collected by Zelditch (1968). Given the importance of language as a carrier of stratification, it is worth noting that a new pronunciation appears to be gaining root to distinguish the two senses. In this new pronunciation, which in the spirit of Humpty-Dumpty may be regarded as the special payment owed multi-meaning words, the “condition” version is pronounced with a short a, as in “map”, and the “prestige” with a long a, as in “day”.¹⁵

2.4. New Immigrant Survey Data

Data are drawn from the first round of the 2003 cohort of the New Immigrant Survey (NIS), a longitudinal study of immigrants admitted to legal permanent residence in different cohorts. The 2003 cohort is the first full cohort to be surveyed; a pilot was carried out on the 1996 cohort. The sampling frame consists of all new LPRs whose records were compiled in the 7-month period May–November 2003. On average, interviews were conducted approximately four months after admission to LPR (mean time elapsed between LPR and interview was 17 weeks and median time was 14 weeks). All respondents were interviewed in the language of their choice; a total of 95 languages were used. The analyses reported in this paper pertain to the Adult Sample, including the main sampled immigrant (N = 8,573), the spouse of the main sampled immigrant (N = 4,334), and a sample of the immigrant's biological children aged 8–12 (N = 1,014). Some of the main sampled immigrants were overseas temporarily at the time of fieldwork (n = 321), and they were administered a short telephone interview. The response rate for the main sampled immigrants in the Adult Sample was 68.6 percent. Appendix Table A.1 reports the basic survey characteristics for the Adult Sample.¹⁶

For each sampled immigrant, the information on the immigrant record in the sampling frame includes the characteristics described in the previous section, for example, whether the immigrant is a new arrival or an adjustee, a principal or an accompanying relative, the type of immigrant visa (for example, spouse of U.S. citizen versus refugee versus diversity), the temporary nonimmigrant visa from which adjustees are adjusting, and the immigrant's country or area of birth.

Immigrants who gain LPR as the spouses of U.S. citizens constitute the largest single category of adult immigrants to the United States, hovering about 33% of all adult LPRs. Meanwhile, employment and diversity principals, in whom there is great interest, comprise far smaller percentages of adult immigrants (5–8% and 4–5%, respectively). Accordingly, the Adult Sample undersampled spouse-of-U.S.-citizen immigrants and oversampled employment and diversity principals. The data include sampling weights, and all percentages and descriptive statistics reported in this paper on based on weighted data (except Appendix Table A.1).

Table 3 summarizes the immigration characteristics of the main adult sampled immigrants and their interviewed spouses. Approximately 89 percent of the adult sampled-immigrant respondents are principals. The spouses of the main adult sampled immigrants include not only principals and nonprincipals, as shown, but also native-born U.S. citizens and previous immigrants (the latter including both LPRs and naturalized citizens). Married couples in which both spouses were interviewed contain one partner from each column. For example, main adult sampled immigrants who are nonprincipals and married are married to spouses who are principals (the second row in the Main Adult column and the first row in the Spouse column); main adult sampled immigrants who are principals and married may have spouses in the third through last rows of the Spouse column (their spouse may be a sponsor, accompanying spouse, contemporaneous legal immigrant, illegal immigrant, etc.).

Table 3.

Immigration Characteristics of Main Sampled Immigrants and Their Spouses Interviewed in the NIS-2003 Adult Sample

Immigration Characteristic	Main Adult	Spouse	Total
Principal	7,615 (88.6%)	714 (15.3%)	8,329
Nonprincipal	958 (11.4%)	NA	958
Native-born U.S. citizen sponsor of spouse	NA	522 (22.3%)	522
Other native-born U.S. citizen	NA	39 (.58%)	39
Foreign-born U.S. citizen sponsor of spouse	NA	566 (24.7%)	566
Other foreign-born U.S. citizen	NA	43 (.72%)	43
Previous immigrant LPR sponsor of spouse	NA	133 (2.84%)	133
Foreign-born spouse of principal who can have accompanying spouse	NA	1,610 (18.3%)	1,610
Foreign-born spouse of principal who cannot have accompanying spouse	NA	707 (15.3%)	707
Total	8,573	4,334	12,907

Notes: Percentages in Main Adult and Spouse columns are based on design weights for main sampled immigrants. The principal is the person who qualifies for the immigrant visa; in some visa categories, visas are also available for the accompanying spouses of principals. Spouses who are principals (n = 714) are married to nonprincipal main sampled immigrants (n = 958); in the NIS all nonprincipal main sampled immigrants are accompanying spouses of principals. Foreign-born spouses of principals who can have accompanying spouses may be accompanying spouses, contemporaneous legal immigrants (e.g., both with employment principal visas), previous immigrants, or unauthorized immigrants. Foreign-born spouses of principals who cannot have accompanying spouses may be contemporaneous legal immigrants (e.g., with visas as parents of U.S. citizens), previous immigrants, or unauthorized immigrants. Spouses of overseas Main Adult respondents were not interviewed; the overseas respondents were 268 principals and 53 nonprincipals (see section 2.4). Thus, for example, the 714 Spouse principals are married to persons from among the 905 non-overseas subset of the 958 Main Adult nonprincipals.

Thus, married couples in the NIS-2003 are of three broad types: (1) principal and sponsor; (2) principal and accompanying spouse; and (3) principal and other. Interviewed spouses of main sampled immigrants appear in Table 3 as follows: in the principal-and-sponsor type of marriage they are in the third, fifth, and seventh rows of Table 3; in the principal-and-accompanying-spouse type of marriage they are in the first and eighth rows; and in the principal-and-other type of marriage they are in the fourth, sixth, and ninth rows. Except for the analyses which focus on couples and parent-child pairs, all analyses in this paper are based on the main sampled immigrants.

Close to 78.7% of the principals are sponsored. Of course, if the principal is sponsored, so is the accompanying spouse, if any. Inspection of the visa categories of the adult sampled immigrants (whether principals or nonprincipals) indicates that close to 77.3% are sponsored.¹⁷

U.S. citizens who sponsor spouses may be native-born (NB) or naturalized foreign-born (FB) former immigrants. NIS data include information on the spouse's nativity. Thus, to test for a variety of effects of the sponsor's nativity in the models below, we separate the spouse-of-U.S.-citizen category into the two subcategories. The respective percentages are 47.4% NB sponsors and 52.6% FB sponsors.

Table 4 provides the visa-category composition of the main sampled immigrants and provides as well, for each visa category, the proportion female and, separately by sex, the average age and schooling and the proportions adjustee and interviewed in English only.

Table 4.

Basic Characteristics of New Legal Immigrants Aged 18+: NIS-2003 Cohort

Immigrant Class of Admission	Percent Female	Age		Schooling		% Adjustees		English Fluency
		Men	Women	Men	Women	Men	Women	Men	Women
Spouse of NB U.S. citizen (16.2%)	59.6	31.6	32.1	13.0	13.8	84.2	81.1	60.8	60.5
Spouse of FB U.S. citizen (17.9%)	66.0	34.2	33.1	12.3	12.5	79.3	65.2	43.4	38.0
Parent of U.S. citizen (11.9%)	66.2	65.5	62.7	8.75	6.93	25.3	33.6	20.8	19.7
Minor child of U.S. citizen (3.38%)	41.9	20.2	20.2	11.5	11.9	46.1	41.4	50.5	46.9
Adult single child of U.S. citizen (3.28%)	54.3	31.6	34.8	12.3	12.3	31.8	33.6	48.9	38.3
Adult married child of U.S. citizen (1.72%)	57.7	40.6	39.9	13.2	12.4	20.4	16.8	48.7	45.5
Spouse of adult child of U.S. citizen (1.51%)	48.1	42.4	37.4	12.9	11.2	8.92	12.9	35.7	25.5
Sibling of U.S. citizen (3.94%)	51.4	48.5	48.2	11.8	11.1	8.97	12.9	35.1	22.7
Spouse of sibling (2.49%)	52.8	50.3	46.2	13.0	10.8	3.98	3.98	37.6	19.6
Spouse of LPR (2.44%)	83.5	43.2	40.2	8.65	7.76	47.7	63.9	16.3	10.6
Child of LPR (2.81%)	49.2	34.3	35.0	11.0	11.1	23.5	19.5	27.7	17.2
Employment principal (6.02%)	32.8	37.3	36.8	15.7	15.2	78.9	55.4	78.6	80.7
Employment spouse (3.63%)	77.1	40.4	35.3	14.7	15.2	56.5	76.4	70.1	76.4
Diversity principal (5.53%)	41.1	32.3	32.8	14.5	14.5	8.45	11.5	52.5	45.4
Diversity spouse (2.58%)	48.7	37.7	34.5	14.6	13.1	5.17	3.55	39.1	38.8
Refugee/asylee/parolee principal (5.35%)	42.8	40.7	38.3	12.8	11.8	100	100	39.9	35.1
Refugee/asylee/parolee spouse (1.22%)	74.8	45.3	43.0	13.3	10.9	100	100	36.5	30.1
Legalization (7.98%)	49.8	38.7	37.9	9.03	8.43	100	100	17.0	9.06
Other (.05%)	---	---	---	---	---	---	---	---	---
All immigrants	56.5	38.7	39.1	12.3	11.6	57.9	57.0	44.7	38.4

Notes: Sample size is 8,573. Estimates based on weighted data. The measure of English fluency requires that the interview was conducted entirely in English.

Figure 1 provides a closer look at age and schooling by presenting the quantile functions associated with their distributions, separately by sex. The quantile functions depict age and schooling as functions of relative rank and thus simultaneously enable a look at both the position (quantity and rank) of particular individuals as well as the whole distribution.

Figure 1.

Quantile Functions of Age and Schooling Distributions, by Sex: NIS 2003 Cohort

The plots for age (panels A and B, Figure 1), together with the underlying data, show that the men's and women's age distributions track each other closely, with women slightly older than men in the region below the percentage rank of 8.4, in the regions bounded by 46.3 and 47.4 and by 51.9 and 98.1, and again in the region above 99.9 percent. Thus, and simplifying greatly, men tend to be slightly older in the bottom half of the distribution and slightly younger in the top half of the distribution. The largest difference occurs at the 89.1 percentile, when men's age is 57.2 and women's age is 59.9 – reflecting the larger proportion female in the parent-of-U.S.-citizen category (Table 4).

The plots for schooling (panels C and D, Figure 1) indicate that men have more schooling than women at every point in the distribution except among the top .2 percent. While 1.7 percent of the men report no schooling whatsoever, 4.13 percent of the women do so. At the other extreme, 13.5 percent of the men report 17+ years of schooling, compared to 10.8 percent of the women. The average (Table 4) is higher for men by about seven-tenths of a year.

The sampled adult immigrants in NIS-2003 were born in 168 countries. Table 5 reports the top ten countries of birth. All are in Asia or the Americas. As shown, Mexico has the largest contingent (17.5%). Thus, the new NIS data on LPRs from Mexico may help remedy the longstanding neglect of Mexican legal migration relative to Mexican illegal migration (Bean and Stevens 2003:44–45).¹⁸

Table 5.

Top Ten Countries of Birth, with Percentage, by Gender: NIS-2003 Cohort

Men	Women	All
Mexico 16.1	Mexico 18.7	Mexico 17.5
India 7.24	India 7.34	India 7.30
El Salvador 6.88	Philippines 6.45	El Salvador 6.11
China 4.90	China 5.55	Philippines 5.47
Philippines 4.20	El Salvador 5.52	China 5.27
Guatemala 2.77	Vietnam 3.59	Vietnam 3.08
Vietnam 2.41	Colombia 2.36	Guatemala 2.43
Dominican Republic 2.17	Dominican Republic 2.34	Dominican Republic 2.27
Cuba 2.13	Guatemala 2.16	Colombia 2.08
Haiti 1.98	Haiti 2.09	Haiti 2.04

Notes: Country of birth constructed from two data series, in the government immigrant record and collected in the survey, with additional information from both the administrative record and the survey used to resolve discrepancies. Percentages based on weighted data.

Previous illegal experience is estimated by combining information from the official administrative record and from the survey, following the procedures in Jasso, Massey, Rosenzweig, and Smith (2008). The main pieces of information based on the immigrant record are: (1) legalization visa (Table 4); (2) nonimmigrant code EWI/WI, UU/UN, or missing; and (3) the Warren (2003, unpubl) measure (nonimmigrant tourist visa B2 and reporting the most recent entry six years or more earlier). The information based on the survey is drawn from questions in the trip history, which ask what kind of documents were used on each trip to the United States.

Table 6 reports the proportions by component of the estimate together with the total combined estimates. As shown, the estimated previous illegal experience based on the immigrant record alone is 35.7 percent; this covers illegal experience immediately before adjusting to LPR. The estimate including the survey measures, which cover earlier spells of illegality, is 39.6 percent. A conservative lower bound, including only respondents with a legalization visa or an EWI/WI nonimmigrant code, would be 11.4 percent.¹⁹

Table 6.

Previous Illegal Experience, by Information Source: NIS-2003 Cohort

Source of Information	Men	Women	All
Immigrant legalization visa	9.23	7.02	7.98
Nonimmigrant code EWI/WI	4.14	3.47	3.76
Nonimmigrant code UU/UN	13.4	11.7	12.4
Nonimmigrant code missing	15.0	14.0	14.4
Nonimmigrant Warren measure	4.96	5.23	5.12
Survey measures	24.5	19.3	21.6
Total based on legalization visa or EWI/WI code	12.9	10.3	11.4
Total based on record alone	37.5	34.4	35.7
Total including survey measures	41.3	38.2	39.6

Notes: The information based on immigrant and nonimmigrant visa is from the official administrative immigrant record. The four components based on nonimmigrant visas are mutually exclusive. The Warren measure refers to having a nonimmigrant tourist visa (B2) and reporting the most recent entry six years or more earlier. The two other components – the immigrant legalization visa and the survey measures – may be combined with any of the other codes, so that, for example, a given respondent may be included in both the legalization visa figure, one of the nonimmigrant visa figures, and the survey measures. Thus, the total estimate is less than the sum of the components. The total estimate may be an underestimate, as it may miss new-arrival immigrants who were already living in the United States illegally and do not report it in the survey measures, as well as persons who had been working without authorization.

Thus, it appears that for approximately 40 percent of the cohort a period of illegality is de facto a stage on the road to legality – notwithstanding popular political images to the contrary. From a stratification perspective, legalizing constitutes massive upward mobility.

The top three countries of birth in the subset with either a legalization immigrant visa or an EWI/WI nonimmigrant code are El Salvador (79.8%), Guatemala (66.7%), and Mexico (18.8%). Including all the information based on the immigrant record yields the same top three countries: El Salvador (89.9%), Guatemala (81.5%), and Mexico (71.3%). Including as well the survey measures preserves the rank ordering and increases the estimates to: El Salvador (92.5%), Guatemala (86.7%), Mexico (77.6).²⁰

A race/ethnicity variable was constructed from the responses to the two standard questions on Hispanic origin and race (used in all NIH-funded research and in other surveys, such as the CPS). The five categories to be used in most of the analyses and their (weighted) percentages are: Hispanic, no race (5.63%), Hispanic white (28.5%), nonHispanic Asian (28.2%), nonHispanic black (10.6%), and nonHispanic white (19.5%). The excluded category contains 7.58% of the sample and consists of other race-ethnicity combinations, including multiple-race and nonHispanic, no-race categories. In the fourth analysis below, all the race-ethnicity combinations will be examined, including, for example, Hispanic blacks.²¹

From the NIS data on religion, we constructed a variable with the following categories and (weighted) percentages: Catholic (41.3%), Orthodox Christian (8.71%), Protestant (16.6%), Muslim (6.96%), Jewish (1.27%), Buddhist (4.25%), Hindu (5.57%), other (1.70%), no religion (12.4%). An additional 1.23% did not provide any information on religion. The excluded category in multivariate analyses is no religion. Because of small sample sizes, in some analyses the Jewish, Buddhist, and Hindu categories are placed in the other-religion category. A similar variable was constructed for childhood religion, based on the question, “What religious tradition, if any, were you raised in?”²²

The NIS survey also asked respondents about their family income when they were age 16, compared to families in the origin country. Five response categories were provided: far below average, below average, average, above average, and far above average. Table 7 reports the percentage distributions, by sex, and Figure 2 graphs the responses. As shown, there is little difference between the two sex-specific distributions. Over half of the respondents reported average family income – 54.2 percent of the women and 51.7 percent of the men. The left tails are fatter than the right, with 28–30 percent reporting below average income versus 18–19 percent reporting above average income.

Table 7.

Parental Location in the Origin Country's Stratification Structure: NIS-2003 Cohort

Family Income at Age 16	Men	Women	All
Far below average	10.1	9.83	9.93
Below average	19.5	18.2	18.8
Average	51.7	54.2	53.1
Above average	15.0	14.1	14.5
Far above average	3.66	3.64	3.65
Number of respondents	4,077	4,374	8.451

Notes: Percentages based on weighted data.

Figure 2.

Parental Location in the Origin Country's Stratification Structure, by Immigrant Sex: NIS-2003 Cohort. The women's distribution has the higher mode.

Legislation creating the Department of Homeland Security (DHS) was enacted in November 2002, and immigrant processing previously carried out by INS transitioned to the new U.S. Citizenship and Immigration Services in March 2003. Thus, many, perhaps most, of the immigrants in the NIS-2003 were processed by both the old INS and the new USCIS. Accordingly, we use “INS” and “USCIS” interchangeably as shorthand for the more precise INS/USCIS.

3. LOST DOCUMENTS, DURATION OF VISA PROCESS, AND LIFE CHANCES

3.1. Lost Documents and the Immigration Context

In the world of U.S. immigration and U.S. travel, two complaints are universal: (1) lack of courtesy among U.S. personnel who deal with the foreign-born – (a) abroad, in the visa sections of embassy and consular posts, administered by the State Department; (b) at ports of entry into the United States, staffed by U.S. Customs and Border Protection, an agency of DHS, since March 2003 and previously by INS; and (c) in the United States, in offices of USCIS and previously in INS offices -- and (2) the propensity of U.S. government immigration agencies to lose documents from case files. There is hardly an academic conference without some story of lack of courtesy or lost documents.

Not surprisingly, lost documents have become a staple of internet immigration blogs and forums. For prospective immigrants to the United States, lost documents are more than an irritation. Lost documents can prolong the visa process, wreak havoc on carefully-made plans for housing (at origin and at destination) or for children's schooling (at origin and at destination) and lengthen the visa process, or, indeed, even jeopardize the entire immigration process. Lost documents contribute to what Levine, Hill, and Warren (1985:3) call the “emotional costs” of migration, epitomizing a corner of the life chances landscape.²³

Of course, government agencies are not unaware of the problem. The official USCIS document, Welcome to the United States: A Guide for New Immigrants (USCIS 2005:14) offers the following advice:

TIP: Keep copies of all forms you send to USCIS and other government offices. When sending documents, do not send originals. Send copies. Sometimes forms get lost. Keeping copies can help avoid problems.

And lost documents have come to the attention of the USCIS Ombudsman (2007:62).

This section examines rates of lost documents by processing venue, whether the immigrant is a new arrival or an adjustee, origin country, visa type, and gender, and reports an initial set of multivariate analyses. A priori, rates and probabilities of lost documents would be expected to differ by venue (due to differences in caseloads and staffing), by visa type (due to the length of time that a case is in the system and thus at risk of having documents lost), by origin country (due not only to venue effects but also to the effect of country on duration of the visa process for numerically limited visas), by previous illegal experience (as, net of visa type, it may trigger further checks and thus affect the length of time that a case is in the system), and by the principal's number of accompanying relatives (as the greater the number of documents in the case, the greater the risk of having a document lost). A priori one might think that given both the popular image and the concern expressed in both A Guide for New Immigrants and the CIS Ombudsman's report, lost documents are more of a problem at INS/USCIS than at the State Department; and certainly differences in workload and agency funding would point in that direction as well.

Beyond such case factors, lost documents may also reflect social mechanisms. For example, State Department personnel, who are living abroad, may be more sympathetic to visa applicants than INS/USCIS personnel in the United States. Further, lost documents may reflect a certain lack of care, a lack of care which may differ systematically across immigrants by their race, gender, religion, or national origin.²⁴

3.2. Empirical Setup and NIS Data for Analysis of Lost Documents

Consistent with the foregoing discussion, we set up an analytic framework in which the case is the unit of analysis (the sample is restricted to principals, who, as shown in Table 3 and discussed above, comprise 89 percent of the main-adult sampled respondents). There are two parallel lines of inquiry. The first assesses the effect of processing venue, distinguishing between cases processed only by the State Department, cases processed only by INS/CIS, and cases processed in both venues, as shown in Table 2. The second examines differences in lost documents across new arrivals and adjustees. All analyses are carried out separately by gender but include as well pooled versions that test the direct effect of gender. Most specifications include country of birth. Specifications that test for discrimination also include ethnicity, represented by race, Hispanic origin, and religion.

The NIS-2003-1 asked a series of questions about the process which led to acquiring the immigrant visa – many of these stimulated by the lively oral tradition concerning aspects of the visa process -- and a randomly selected half of the main adult respondents were asked, “Were any documents or files lost during the process?” This is the question analyzed in this section.²⁵

3.3. Migration and Stratification: Lost Documents

Table 8 reports the proportion of principals in whose cases documents were lost, by processing venue. As shown, the overall rate was 11.3 percent, with slightly more male principals than female principals experiencing lost documents – 12.2 versus 10.6 percent. There is large variation by processing venue, a variation which mimics caseload. Cases processed by the State Department alone total only 5.7 percent of the cases, and in this set the proportion with lost documents is 3.52 percent. At the other extreme, cases processed by INS/USCIS alone total 55 percent, and the proportion with lost documents is 13.5 percent. Finally, cases processed by both State and INS/USCIS total 39.3 percent, and, as would be expected, have a larger proportion with lost documents than those processed by State alone but a smaller proportion than those processed by INS/USCIS alone – 9.27 percent. Thus, it seems clear that the a priori conjectures are correct and that documents are more likely to be lost in INS/USCIS offices than in consular posts overseas or the U.S. facilities operated by the State Department (notably the National Visa Center and the Kentucky Consular Center). Of course, this may reflect the type of cases processed at the different venues, in particular, cases involving principals with previous illegal experience, a possibility examined in multivariate analyses below.

Table 8.

Documents Lost During the Visa Process, by Processing Venue and Gender: NIS-2003 Cohort

Processing Venue	Men		Women		All
	Percent in Sample	Percent Lost Docs	Percent in Sample	Percent Lost Docs	Percent in Sample	Percent Lost Docs
State only	7.54	4.37	4.28	2.36	5.70	3.52
INS/CIS only	58.4	14.7	52.4	12.5	55.0	13.5
Both State and INS/CIS	34.1	9.54	43.3	9.11	39.3	9.27
Total lost documents	NA	12.2	NA	10.6	NA	11.3
Number of observations	1853	NA	1930	NA	3783	NA

Notes: NA = not applicable. See Table 2 for definition of processing venue. Percentages based on data weighted to adjust for sampling stratification.

Table 9 shifts perspective, reporting rates of lost documents separately for new arrivals and adjustees and providing a look by country and visa type. Adjustees are substantially more likely to experience lost documents than new arrivals – 13.7 versus 7.56 percent (panel C).

Table 9.

Immigrant Principals Whose Documents Were Lost During the Visa Process: By Origin Country, Visa Category, and Sex: NIS-2003 Cohort

New Arrivals		Adjustees
Men	Women	Men	Women
A. Percent Documents Lost in Top Ten Immigration Countries
Dominican Republic 12.4	Dominican Republic 11.0	China 16.3	Mexico 14.2
Mexico 10.0	Philippines 9.54	Mexico 15.5	Guatemala 10.5
Vietnam 9.14	Haiti 8.64	Cuba 15.4	India 9.78
China 4.97	Vietnam 6.11	Philippines 11.4	El Salvador 9.24
Philippines 3.99	India 4.84	India 10.2	Cuba 8.79
India 2.56	Mexico 3.06	El Salvador 8.31	Philippines 3.47
---	China 1.66	Guatemala 3.74	China .897
B. Visa Categories with Five Highest Rates of Lost Documents
Married Child of Cit 13.9	Spouse of LPR 15.4	Spouse of FB Cit 20.1	Single Child of Cit 25.7
Single Child of Cit 11.5	Employment 14.8	Child of US Cit 16.7	Diversity 21.8
Spouse of FB Cit 10.9	Spouse of NB Cit 13.8	Spouse of NB Cit 16.0	Employment 19.1
Sibling of US Cit 9.34	Spouse of FB Cit 11.0	Ref/Asy/Par 14.3	Spouse of FB Cit 15.2
Employment 9.18	Child of LPR 8.14	Employment 13.9	Spouse of NB Cit 12.0
C. Overall Proportion with Documents Lost – All Principals
7.99 (822)	7.25 (937)	14.7 (1031)	12.9 (993)
7.56 (1759)		13.7 (2024)

Notes: The question on documents lost during the visa process was asked of half the main sampled immigrants, randomly chosen. Each principal represents a case in the visa process. New arrivals are processed by consular offices of the State Department overseas; adjustees are processed in the United States by offices of the Immigration and Naturalization Service (until March 2003) and the U.S. Citizenship and Immigration Services (starting in March 2003). Proportions are based on weighted data, in cells with at least 20 observations. Overall proportions with documents lost appear in panel C, with the number of observations in parentheses under the overall proportions.

Of the 168 countries of birth represented in the Adult Sample, 147 are represented in cases involving principals in the subsample asked the lost documents question. Among these 147 countries, 65 have no emigrants reporting lost documents. More sharply, within the four subsets defined by gender and adjustee status, there are no cases of lost documents for 76 out of 104 countries among new-arrival men, 68 out of 103 among new-arrival women, 64 out of 116 among adjustee men, and 49 out of 102 among adjustee women. Thus, lost documents appear not to be a universal phenomenon.

Table 9, panel A, reports the lost document rates for the top ten origin countries (Table 5). These figures underscore the variation in lost documents across country. For example, among adjustee women from the top ten origin countries, the rates vary from less than 1% for those born in China to more than 14% for those born in Mexico; among new arrivals, the rates vary from lows of 1.66% and 2.56% among women from China and men from India, respectively, to over 11% for both men and women from the Dominican Republic. Gender appears to be a factor; contrast the rates of lost documents among adjustees from China; women, as noted, have the lowest rate, but men the highest – 16.3 percent. Again, these country and gender effects invite multivariate scrutiny.

Visa category shows even greater variation than country of birth (Table 9, panel B). The largest rates are for adjustee adult unmarried daughters of U.S. citizens, over a quarter of whom experience lost documents. The lowest rates (not shown) are among new-arrival women with diversity and married-daughter-of-U.S.-citizen visas – 2.36 and 2.34 percent, respectively.

A result which catches the eye is that in three of the four subsets in Table 9, spouses of foreign-born U.S. citizens have higher rates of lost documents than spouses of native-born U.S. citizens – for example, 20.1 versus 16 percent and 15.2 versus 12 percent among adjustee husbands and wives of U.S. citizens, respectively. The exception is among new-arrival wives – wives of native-born U.S. citizens have a lost document rate of 13.8 percent versus 11 percent among wives of foreign-born U.S. citizens. As with origin country, visa type also invites multivariate scrutiny.

Table 10 reports the results of two sets of multivariate analysis, designed to more sharply assess the processing venue and adjustee effects. Each set includes three specifications for each sex. The first two are binary logit, with robust standard errors. Specification (3) is a fixed-effects logit, with fixed effects for the full set of countries. Specifications (2) and (3) also include previous illegal experience. The processing venue results, reported in panel A, indicate that processing venue is statistically significant in all specifications, and the effect is exactly in line with the raw rates. Cases processed only by INS/USCIS have the highest probability of lost documents, followed by cases processed by both INS/USCIS and State, and cases processed by State only have the lowest probability. Previous illegal experience increases the probability of lost documents but is statistically significant only in the women's fixed-effects logit specification.²⁶

Table 10.

Logit Estimates of Venue and Adjustee Effects on Documents Lost During the Visa Process, by Sex: NIS-2003 Cohort

Regressor	Men			Women
	(1)	(2)	(3)	(1)	(2)	(3)
A. Which processing venue has the higher probability of lost documents?
INS/CIS only	1.301	1.150	1.480	1.906	1.707	1.309
Both State and INS/CIS	.905	.865	1.096	1.584	1.523	1.150
Joint test chi2 (2 df)	21.52***	13.73***	14 95***	19.05***	12.72**	6.42*
Illegal experience	NS	.277 (1.46)	.302 (1.43)	NS	.303 (1.50)	.571** (2.74)
B. Do new arrivals or adjustees face the higher probability of lost documents?
Adjustee	.550 (2.59)**	.319 (1.36)	.307 (1.18)	.645 (3.10)**	.471 (2.05)*	.517 (2.17)*
Spouse of NB citizen	.754	.750	.950	.275	−.410	−.313
Spouse of FB citizen	1.115	1.122	1.500	.525	−.144	.227
Parent of US citizen	.433	.453	.735	−.340	−1.018	−.777
Child of US citizen	.774	.790	.853	.625	−.0669	−.0585
Adult single child of U.S. citizen	1.075	1.088	1.175	.522	−.166	.00255
Married child of US cit	1.280	1.311	1.217	---	−.692	−.581
Sibling of US cit	.855	.890	1.358	−.513	−1.186	−.653
Spouse of LPR	---	---	---	.264	−.480	−.174
Child of LPR	.666	.702	.812	.694	---	---
Employment	.710	.891	1.250	.818	.195	.278
Diversity	−.0575	.0228	.0175	−.440	−1.093	−.420
Refugee/asylee/parolee	.666	.749	.811	.00162	−.629	−.249
Legalization	.103	−.0107	.657	−.102	−.887	−.334
Joint test chi2 (12 df)	19.20	20.56	15.42	29.82**	31.47**	16.86
Illegal experience	NS	.431* (2.15)	.360 (1.55)	NS	.309 (1.54)	.408 (1.78)
Number of observations	1853	1853	1597	1930	1930	1692

Notes: NS = not in specification. Specifications (1) and (2) are binary logit, with robust standard errors. Specifications (3) are fixed effects logit, with fixed effects for the full set of countries. Absolute values of asymptotic t-ratio appear in parentheses under parameter estimates for numeric and binary variables. Joint tests reported for multiple-category categorical variables.

^*p < .05,

^**p < 0.1,

^***p <. 001

two-tailed tests for single coefficients, one-tailed for joint tests

Versions pooled by sex did not find a statistically significant effect of sex, although the signs indicate that men are more likely to experience lost documents than women. Additional versions, incorporating race and Hispanic origin, as well as religion, did not find religion effects; the only notable effect is that among women, nonHispanic blacks have a substantially lower probability of having lost documents than immigrants of other race and Hispanic origin combinations.

Panel B reports the results for the specifications assessing the adjustee effect. This set also incorporates visa type. In all specifications, the adjustee effect is positive, and it achieves statistical significance in all the women's specifications and one of the men's. Previous illegal experience is also uniformly positive, but statistically significant only in one of the men's specifications. Visa type is statistically significant in two of the women's equations. The point estimates indicate that in all the specifications, spouses of foreign-born U.S. citizens have higher probability of lost documents than spouses of native-born U.S. citizens. In the women's equations, employment principals have the highest probability of lost documents, but the documents could have been lost in the Department of Labor. Parents of U.S. citizens have among the lowest probabilities of lost documents.

As with the venue regressions, we tested for effects of gender, race and Hispanic origin, and religion (not shown). None were statistically significant. The effect of gender was close to zero.

3.4. Aftermath of Lost Documents: Lengthening the Visa Process for Adjustees

Overall, documents were lost in 11.3 percent of the immigration cases in the NIS-2003 sample. Are there negative consequences for the new immigrants? Was it a minor irritation, or will there be lasting consequences? Because the NIS is a longitudinal study, future rounds of the survey will make it possible to examine the long-term effects, if any, of having documents lost – including diminished attachment to the United States, visible in naturalization, emigration, and voting. For now, we focus on two more immediate consequences, both of which could also have longer-lasting effects of their own: lengthening the processing period and becoming depressed. Visa depression will be analyzed in the next section. Here we concentrate on the consequences of lost documents for the length of the visa process.

We begin with a look at processing time. As already noted, the visa process lasts from the filing of the first document to granting of legal permanent residence. A priori there are several mechanisms affecting duration of the visa process, some of which work at cross purposes. First, the visa process should be longer for numerically limited visas which are backlogged – in 2003 these were family preference visas (State Dept, Visa Bulletins). Other visas are not subject to waiting for a visa number; moreover, diversity visas must be processed within the fiscal year. Second, the visa process should be longer for adjustment of status cases than for new-arrival cases, because the volume is larger stateside (Tables 2 and 8) and the per-case resources appear to be lower than in U.S. consulates abroad. Third, however, the visa process should be longer for new-arrival cases because, while among adjustees approval leads immediately to LPR (indicated by a stamp in the passport), among new arrivals that same approval is only the first of two approvals, yielding a visa which is valid for six months as the prospective immigrant prepares to travel to the United States, where a U.S. agent conducts an inspection and provides the second approval, authorizing admission to LPR (again indicated by a stamp in the passport). Fourth, within visa categories that provide visas for both principals and accompanying spouses, new arrivals granted LPR as spouses of principals should have a shorter visa process than principals because the marriage might have occurred after the initial petition was filed. Fifth, however, new arrivals granted LPR as spouses of principals should have a longer visa process because they are allowed an additional six months for “following to join” the principal. Sixth, employment cases requiring labor certification (second and third preference categories) should have a longer visa process than cases not requiring it. Finally, country of birth also affects duration of the visa process in the numerically limited preference categories.²⁷

Note that the second and third mechanisms have opposite effects, as do the fourth and fifth. Which mechanism is stronger is an empirical question to be examined below.

Table 11 reports the duration of the visa process in the NIS-2003 cohort, separately for new arrivals and adjustees, for principals and spouses, and by gender. The first result which hits the eye is a result not anticipated from the mechanisms listed above: In each of the four subsets, visa processing takes longer for spouses of foreign-born U.S. citizens than for spouses of native-born U.S. citizens. The reason is not immediately obvious. Inspection of the requisite Form I-130 (“Petition for an Alien Relative”) which must be filed by the sponsor indicates that the only difference between the two types of sponsors pertains to the evidence of their citizenship that must be presented, namely, while both native-born and foreign-born citizens can present a passport, other evidence includes a birth certificate for a native-born citizen and a certificate of naturalization (or of citizenship) for a foreign-born citizen. Thus, there are two further avenues to explore: (1) whether marriage cases involving foreign-born U.S. citizens are more complicated in an immigration sense (i.e., are higher-order marriages for one or both spouses, the sponsored spouse has difficulty accessing the requisite documents, such as military and police records, or the documents have to be translated from a non-Roman alphabet, etc.), and (2) whether marriage cases involving foreign-born U.S. citizens receive greater scrutiny from U.S. officials. Both are outside the scope of this paper. But note that the NIS data provide sufficient information to establish differences between the two types of marriage cases that may affect processing times. Note also that the longer visa process for spouses of foreign-born U.S. citizens than for spouses of native-born U.S. citizens may reflect the effect of lost documents, to be examined below.

Table 11.

Average Years of Immigrant Visa Processing Time, by Visa Characteristics and Sex: NIS-2003 Cohort

	New Arrival				Adjustee
Immigrant Class of Admission	Principal		Spouse		Principal		Spouse
	Men	Women	Men	Women	Men	Women	Men	Women
Spouse of NB U.S. citizen	1.23	1.11	NA	NA	2.39	2.15	NA	NA
Spouse of FB U.S. citizen	1.88	1.88	NA	NA	3.60	2.78	NA	NA
Parent of U.S. citizen	2.65	2.38	NA	NA	2.54	2.58	NA	NA
Minor child of U.S. citizen	2.80	3.74	NA	NA	4.78	4.59	NA	NA
Adult single child of U.S. citizen	6.96	6.41	NA	NA	8.17	9.06	NA	NA
Adult married child of U.S. citizen	7.71	8.19	6.45	6.69	8.93	8.84	---	---
Sibling of U.S. citizen	13.2	13.7	12.4	12.7	10.39	13.8	---	---
Spouse of LPR	6.34	7.70	NA	NA	7.73	8.54	NA	NA
Child of LPR	9.09	8.85	NA	NA	11.5	11.3	NA	NA
Employment	2.67	2.25	2.50	3.76	4.04	4.73	4.48	3.36
Diversity	2.01	2.27	2.32	2.29	2.15	2.04	---	---
Refugee/asylee/parolee	NA	NA	NA	NA	5.82	5.45	6.57	6.43
Legalization	NA	NA	NA	NA	7.08	5.95	NA	NA
Other	---	---	---	---	---	---	---	---
All immigrants	4.58	4.18	6.48	6.59	4.89	3.90	5.80	4.59

Notes: Sample size is 8,573. Estimates based on weighted data. Combinations which either do not arise in immigration law or do not appear in the sample are denoted NA. The mark “—” indicates cells with observations fewer than 20.

Other results illuminate the mechanisms described above. As expected, numerically unlimited cases (spouses, parents, and minor children of U.S. citizens), diversity cases, and employment cases have the shortest visa process. Sibling cases have the longest visa process.

Contrasting adjustee and new-arrival visa process times within subsets of principals indicates that in almost every visa type, the adjustee process is longer than the new-arrival process, suggesting that the agency mechanism trumps the behavioral mechanism (new arrivals taking up to six months to settle affairs before traveling to the United States). For example, the visa process for spouses of native-born U.S. citizens lasts 1.23 and 1.1 years, on average, for men and women, respectively, who are new arrivals, but almost twice as long for adjustees – 2.39 and 2.15 years for men and women, respectively. These figures also provide an empirical grounding for the perennial discussion among visa applicants and immigration lawyers concerning the relative merits of adjustment of status and consular processing (a search of immigration forums and chatrooms on the internet will quickly yield pertinent anecdotes), as well as to the policy of permitting employment-based visa applicants residing in the United States to choose consular processing (as shown in Form I-140, “Petition for Alien Worker”).

Within new arrivals, spouses of principals have shorter visa process than principals among numerically limited married children and siblings of U.S. citizens – indicating that they may have married after the principal entered the visa queue. Differences in duration of the visa process between principals and spouses are trivial among employment and diversity immigrants, except among employment new arrival women, who exhibit the opposite pattern – longer visa process for spouses of principals – presumably because the visa wait is shorter and the spouses follow later.

For visual illustration, Figure 3 provides the quantile functions of the duration distributions for the two extremes of the visa process – among spouses of native-born U.S. citizens and siblings of U.S. citizens – separately for new arrivals and adjustees and by gender. The plots for the spouses vividly show the longer duration among adjustees than among new arrivals. The expected discrepancy between new arrivals and adjustees among siblings is more ambiguous, possibly because the arrival-adjustee subsets contain a different origin-country mix, and the wait for these numerically limited visas differs by country.

Figure 3.

Two Extremes of Duration of Visa Process: Quantile Function of Duration Distribution Among Immigrant Spouses of Native-Born U.S. Citizens and Siblings of U.S. Citizens in the NIS-2003 Cohort. Panels A and C depict new arrivals, and Panels B and D depict adjustees. Each grid includes both the men's and women's distribution.

But the main question in this section concerns the effect of lost documents on duration of the visa process. Table 12 reports OLS estimates of the effect of lost documents, controlling for visa type (which as expected from immigration law and as documented in both Table 11 and Table 12 has its own effect). Having documents lost has a statistically significant effect on the duration of the stateside visa process, prolonging it by one year, on average, for men adjustees and by almost eleven months for women adjustees (coefficients of 1.017 and .899, respectively). Among new arrivals, the effect of lost documents does not reach statistical significance for either sex, and the point estimates indicate a lengthening of the visa processing period by about half a year for men and nothing for women (coefficients of .572 and −.018, respectively). The reason for the venue differential in the effect of lost documents on the length of the visa-processing period may be that documents are easier to replace in the origin country than in the United States.

Table 12.

OLS Estimates of Effects of Lost Documents and Visa Characteristics on Duration of Visa Process, by Sex: NIS-2003 Cohort

Regressor	New Arrivals		Adjustees
	Men	Women	Men	Women
Documents lost	0.572 (1.42)	−0.0180 (.05)	1.017 (2.89)**	0.899 (3.07)**
Spouse of NB citizen	−8.428	−8.104	1.035	.398
Spouse of FB citizen	−8.073	−7.255	1.829	1.090
Parent of U.S. citizen	−6.855	−6.819	.622	.919
Child of U.S. citizen	−6.905	−6.287	3.613	2.705
Adult single child of U.S. citizen	−2.337	−2.480	5.733	6.244
Adult married child of U.S. citizen	−2.263	−.990	7.377	5.910
Spouse of adult child of U.S. cit	−2.831	−.945	1.185	8.233
Sibling of U.S. citizen	3.315	4.774	9.280	13.289
Spouse of sibling of U.S. citizen	2.564	3.131	13.351	11.542
Spouse of LPR	−3.776	−1.563	6.178	6.575
Child of LPR	---	---	8.310	10.166
Employment principal	−7.227	−6.670	2.729	2.989
Spouse of employment principal	−7.213	−5.738	2.656	1.893
Diversity principal	−7.972	−6.789	.683	.175
Spouse of diversity principal	−6.709	−6.937	---	---
Refugee/asylee/parolee principal	NA	NA	4.099	3.626
Spouse of refugee/ asylee/parolee principal	NA	NA	5.027	5.582
Legalization principal	NA	NA	4.716	4.025
Joint test F	81.1*** (14, 898 df)	51.2*** (14, 983 df)	26.75*** (17, 2004 df)	79.15*** (17, 1043 df)
Adjustee, EWI/WI	NA	NA	.521	1.810
Adjustee, UU/UN	NA	NA	1.038	.718
Adjustee, no code	NA	NA	.868	.789
Adjustee, Warren overstay	NA	NA	.444	1.193
Joint test F	NA	NA	3.10* (4, 1004 df)	6.01*** (4, 1043 df)
Constant	9.761 (23.41)***	9.103 (16.68)***	1.052 (2.02)*	1.116 (1.53)
R-squared	.593	.591	.240	.367
Observations	914	999	1027	1066

Notes: The visa process lasts from the date of filing the first application to the date of admission to legal permanent residence. NA = not applicable. Absolute values of asymptotic t-ratio appear in parentheses under parameter estimates for numeric variables. Joint tests reported for multiple-category categorical variables.

^*p < .05,

^**p < .01,

^***p < .001

two-tailed tests for single coefficients, one-tailed for joint tests

Table 12 shows that the pattern of longer visa process for spouses of foreign-born U.S. citizens than for spouses of native-born U.S. citizens persists after controlling for lost documents. Table 12 also shows that, as expected, among adjustees, those immigrants whose official records indicate previous illegal experience had a longer visa process. Sharp understanding of the patterns revealed in the coefficients, however, requires further research.²⁸

4. VISA DEPRESSION AND LIFE CHANCES

The “emotional costs” of applying for an immigrant visa have long been noted (Levine, Hill, and Warren 1985:3). Both the outcome and its timing are uncertain; moreover, as we have seen, documents may be lost. NIS data enable examination of an extreme form of some of these emotional costs, namely, experiencing sadness and depression due to the process of applying for a visa – visa stress. Approximately 17% of the new immigrants report experiencing visa depression. This section explores risk factors and protective factors, focusing in particular on links with race and gender, visa category and previous illegal experience, and religion. As well, this section provides estimates of the effect of having documents lost on visa depression.

A priori, we expect visa stress to differ by conditions and characteristics associated with the visa process. Diversity immigrants are expected to have less visa stress, in part because, as noted above, the entire process must be concluded in a relatively short period of time (section 2.2 above). Refugees and asylees, too, are expected to have less visa stress in the period just before admission to LPR because for them the daunting and uncertain part was obtaining the initial refugee or asylee status, with adjustment to LPR being more routine (sections 2.2 and 2.3). However, as noted above, adjustees are expected to have more visa stress than new arrivals because, following the logic of Simmons and Blyth's (1987) argument concerning puberty and school transitions, they undergo both the visa process and the process of adjusting to the United States at the same time (section 2.3).

4.1. Empirical Setup and NIS Data for Analysis of Visa Depression

Two sets of analyses are carried out, the first on the full sample of the NIS-2003 main sampled immigrants (Table 3), the second on the half-sample who were asked the lost documents question.

The NIS-2003-1 (attentive to the speculation in the oral tradition) asked all main sampled immigrants except the 321 overseas respondents (section 2.4 and Appendix Table A.1) the question, “During the past 12 months, have you ever felt sad, blue, or depressed because of the process of becoming a permanent resident alien?” Approximately 17.4 percent of the new immigrants reported becoming depressed due to the visa process. Moderately more women than men became depressed – 18.5% of the women versus 15.9% of the men – and moderately more adjustees than new arrivals became depressed – 18.7% of adjustees versus 15.5% of new arrivals.

The other new variable introduced in the analysis of visa depression pertains to having suffered harm. A question in the NIS-2003-1 asked, “Did you or your immediate family ever suffer any harm outside of the United States because of your political or religious beliefs, or your race, ethnicity, or gender?” Approximately 7 percent of the sample answered yes. These respondents are concentrated in the refugee/asylee/parolee categories – 46% and 49% of principals and spouses, respectively – together with 14.9% of legalization immigrants. Within these visa categories, there is a further concentration by origin country; in the RAP categories 12.2% of those who suffered harm are from Ukraine and 11.9% from Russia, and in the legalization category 66% are from El Salvador and 23% from Guatemala.

4.2. Migration and Stratification: Visa Depression

Table 13 reports conditional logit estimates of the visa depression equation with country-of-birth fixed effects. As expected from the overall percentages, the pooled versions (not shown) indicate that women are statistically significantly more likely to experience visa depression than men (with prob values of .003 in the full sample and .005 in the subsample). Visa category and the adjustment variable are statistically significant in the men's full-sample equation (beyond the .004 level and the .000 level, respectively) but neither is significant in the women's equations (and only adjustee reaches significance in the men's subsample equation).

Table 13.

Logit Estimates of Depression Due to the Visa Process, by Sex: NIS-2003 Cohort

Regressor	Full Sample		Subsample with Data on Lost Documents
	Men	Women	Men	Women
Age	0.0324	0.0513	0.0224	0.0692
Age squared	−0.000446	−0.000668	−0.000279	−0.000834
Joint test chi2 (2 df)	2.92	8.85*	0.47	5.67
Schooling	0.00594 (0.42)	0.0167 (1.39)	0.0404 (1.92)	0.0268 (1.55)
Spouse of NB citizen	−0.115	0.232	−0.271	0.158
Spouse of FB citizen	−0.288	−0.216	−0.807	−0.193
Parent of U.S. citizen	−0.213	−0.0108	−0.694	−0.0714
Child of U.S. citizen	−0.345	0.344	−0.227	0.155
Sibling of U.S. citizen	−0.0375	−0.381	0.278	−0.384
Spouse of sibling of U.S. citizen	−0.0919	−0.118	0.0189	−0.862
Employment principal	0.0765	0.0445	−0.183	−0.0215
Spouse of employment principal	0.0609	−0.195	−0.226	−0.643
Diversity principal	−0.352	−0.428	−0.716	−0.253
Spouse of diversity principal	−0.713	−0.519	−1.578	−0.528
Refugee/asylee/parolee principal	−1.149	0.104	−1.203	−0.243
Spouse of refugee/ asylee/parolee principal	−0.586	−0.273	−2.007	−0.277
Legalization principal	0.536	−0.00346	0.284	−0.497
Joint test chi2 (13 df)	30.54**	17.03	20.68	11.42
Adjustee	0.591 (3.75)***	−0.0358 (0.26)	0.459 (2.01)*	−0.0259 (0.13)
Adjusting from illegality	−0.173 (1.11)	0.00821 (0.06)	−0.190 (0.82)	0.122 (0.61)
Experienced harm	0.409 (2.34)**	0.238 (1.23)	0.403 (1.43)	0.292 (0.99)
Documents lost	NA	NA	0.770 (4.15)***	0.468 (2.60)**
Log likelihood	−1504.27	−1703.95	−657.86	−800.53
Observations	3825	4070	1812	1969

Notes: NA = not applicable. Specifications include full set of country fixed effects. Absolute values of asymptotic t-ratio appear in parentheses under parameter estimates for numeric and binary variables. Joint tests reported for multiple-category categorical variables.

^*p < .05,

^**p <.01,

^***p <.001

two-tailed tests for single coefficients, one-tailed for joint tests

The question whether having an employer or a relative in the United States – perhaps a special kind of relative – confers protection against depression or instead operates to promote stress receives some hints from the statistically significant results for men and the point estimates for women. Among immigrants married to (and sponsored by) U.S. citizens, those married to foreign-born U.S. citizens appear to gain more protection against depression than those married to native-born U.S. citizens; this advantage is particularly visible in the women's subsample equation, where wives of native-born U.S. citizens have the second-highest probability of visa depression, while wives of foreign-born U.S. citizens rank tenth. Young men (with a child-of-U.S.-citizen visa) also appear to benefit from having a U.S. citizen parent, but young women with a U.S. citizen parent have the highest probability of visa depression in the full sample and the second-highest in the subsample. These results provide further hints of the gender-specific character of social capital and of the degrees of social capital (Curran, Garip, Chung, and Tangchonlatip 2005).

As expected, diversity lottery immigrants of both sexes and both principals and spouses appear to have moderately low probability of visa depression, as do refugees/asylees/parolees and their spouses, except for RAP principal women (full sample). Employment-based immigrant men have some of the highest probabilities of visa depression -- with principals and spouses ranking second and third, respectively, in the full sample – while wives of employment principals rank 9^th and 13^th in the full sample and subsample, respectively.

As noted, the adjustee variable is statistically significant and positive in the men's equations but negative and not reliably estimated in the women's equations. To test whether the effect is of adjusting per se or of adjusting from illegality, the equations include a variable for adjusting from illegality. Though not reaching statistical significance, the coefficient in the men's equation is negative, indicating that adjustees who are adjusting from illegality have a lower probability of visa depression than adjustees who are adjusting from a valid nonimmigrant visa. Moreover, like the adjustee variable, the variable for adjusting from illegality it hints at diametrically opposite patterns by gender.

Overall, then, the coveted employment visas confer no protection from depression on principals, but do appear to protect their wives. Refugees are a highly select set of survivors, and it is likely that the hardships of the visa process pale next to the hardships that made them refugees – and the earlier process by which they acquired the initial refugee or asylee status. Diversity immigrants -- winners of a lottery -- may be basking in the glow of fortune, not to mention the expedited handling of their visas, given, as noted, legal requirements to admit to LPR status within the fiscal year. And not all U.S. citizen sponsors of spouses are the same. The foreign-born among them, whose spouses' immigration cases have greater probability of lost documents and longer duration, appear to do more to prevent or mitigate depression among their spouses, with the discrepancy larger among cases involving U.S. citizen husbands and immigrant wives than among cases involving U.S. citizen wives and immigrant husbands.

Among personal characteristics, age matters for women. The probability of depression increases with age until the late thirties, and then declines.

Schooling appears not to protect against depression; the coefficients do not reach significance, although their signs are uniformly positive. Having experienced harm increases the probability of depression, statistically significantly so for men in the full sample. Thus, refugees who suffered harm have both the protection attached to refugees and the risk attached to harm.

Turning to the effect of having documents lost during the visa process, the subsample equations indicate that having documents lost significantly increases the probability of visa depression for both men and women.

Thus, adjustees would seem to endure a “triple whammy.” First, they are more likely to have documents lost. Second, if they have documents lost, processing time is lengthened by about a year (perhaps because the documents are more difficult to replace in the United States than in the origin country). Third, holding constant lost documents, adjustee men have a higher probability of becoming depressed during the visa process.

These results suggest that the hardships of the visa process are more difficult to endure if they occur at the same time that immigrants are trying to adjust to life in a new country. The “comforts of home” would seem to mitigate visa stress.

Finally, as with lost documents, the longitudinal nature of the New Immigrant Survey will make it possible to gauge the long-term consequences of having experienced visa depression, both with respect to health and with respect to the degree of attachment to the United States.²⁹

5. PRESENTATION OF SELF: GENDER, STATUS, AND DECLARING ONESELF THE PRINCIPAL

As discussed above, the immigrant principal is the individual who qualifies for the immigrant visa (section 2.2). In many visa categories – obviously excepting spouse principals (spouses of U.S. citizens and of LPRs) and unmarried children of U.S. citizens and of LPRs but also excepting parents of U.S. citizens – immigrant visas are also made available to the spouse of the principal. Being a principal appears to confer a certain social status. Anecdotal evidence suggests that immigrants who are actually spouses of principals may like to be thought of as principals. After all, being a principal announces to the world that one has qualified for a coveted immigrant visa and, moreover, except for spouses sponsored by U.S. citizens and LPRs, that one is not indebted to one's husband or wife or to one's in-laws for the immigrant visa. Thus, in a Goffman (1959) sense, declaring oneself the principal is a perfect ingredient in the presentation of self.

A notable possible exception to this status interpretation of being a principal pertains to diversity visas, which are awarded by lottery; it is thought to be quite common for married couples (in which both spouses satisfy the schooling or experience requirement) to submit two separate applications, with each spouse as principal, and fortune picks the winner. Thus, there being no special merit in winning a diversity visa – beyond the favor of the gods – the phenomenon of declaring oneself the principal may be muted in diversity visas, unless luck is thought to signal other good qualities or even the promise of further luck.

Meanwhile, as we have seen, the reality of applying for an immigrant visa is, for most applicants, a test of enduring adversity and disrespect. Moreover, social status may decline in the United States (Bean and Stevens 2003:28). The stage is set for the right salve. In the world of immigration, declaring oneself the principal – what we may call the “me-principal” assertion -- provides a perfect mix. Perhaps a little lie, but a harmless one and one with no apparent repercussions.

Concomitantly, however, there may be persons of such modesty and self-effacement – or obsequiousness -- that they are loath to declare themselves the principal, generally avoiding the subject or even misrepresenting the part they played in their family's immigration. Moreover, depression may intensify the self-effacement and lack of assertiveness.

Alongside these mechanisms for protecting the self, there is, of course, the effect of knowledge. Some new immigrants may know more about the visa process than others.

Finally, the immigrant may think that in a just world (s)he would (or would not) be the principal, and attempt to redress the injustice by reporting the just situation rather than the actual situation. As Blanche Dubois puts it in Tennessee Williams' A Streetcar Named Desire, “I don't tell truths. I tell what ought to be truth.”

5.1. Empirical Setup and NIS Data for Analyzing the Me-Principal Assertion

The objective of this analysis is to assess the determinants of declaring oneself the principal. Accordingly, we divide the sample into two subsamples, the first consisting of principals and the second of nonprincipals (in the NIS data, all the Adult Sample main-sampled nonprincipals are accompanying spouses). Using sampling weights, 88.6% of the sample are principals (Table 3).

The equation specification includes the immigrant's age, schooling, and visa characteristics, two childhood variables – childhood religion and parental relative income at age 16 -- as well as visa depression. Interpretation of the estimates differs across subsample. In the subsample of principals, positive coefficients indicate having more information and/or overcoming modesty to assert one's true principalhood. In the subsample of nonprincipals, positive coefficients indicate having incorrect information and/or using the me-principal salve. For example, if schooling increases the likelihood of having correct information or promotes telling the truth, then the schooling coefficient should be positive in the principals equation and negative in the nonprincipals equation.

The NIS-2003-1 (again, building on the oral tradition) asked all main sampled respondents (except the 321 who were overseas – see section 2.4 and Appendix Table A.1), “Did you obtain legal permanent residence because you yourself qualified for an immigrant visa, or because you are the accompanying spouse or child of another immigrant? That is, were you the `principal' immigrant or accompanying the principal immigrant?” Among the true principals, only 78.9 percent responded in the affirmative, while among the nonprincipals, 26.5 percent said yes. Looking at the gender breakdown, men were more likely to declare themselves the principal, whether or not they were – 83.0% of male principals versus 75.6% of female principals and 33.3% of male nonprincipals versus 22.2% of female nonprincipals.

5.2. Migration and Stratification: the Me-Principal Assertion

Table 14 reports binary logit estimates of the me-principal equation, separately for principals and nonprincipals and by gender. As expected from the overall gender percentages, specifications pooled by sex (not shown) indicate that men are statistically significantly more likely to declare themselves the principal, among both principals and nonprincipals (significant beyond the .000 level among principals and beyond the .05 level in the smaller nonprincipals subsample), suggesting a certain male sense of entitlement and/or greater need of a status boost. This male propensity to claim principal status is net of all the other characteristics included in the equation -- net of visa, origin area, and visa depression.

Table 14.

Declaring Oneself the Principal: Logit Estimates, Separately for Principals and Nonprincipals and by Sex, NIS-2003 Cohort

Regressor	Principals		Accompanying Spouses
	Men	Women	Men	Women
Age	0.124	0.509	0.0696	−0.0351
Age squared	−.00109	−.000323	−0.00117	0.000655
Joint test chi2 (2 df)	27.98***	20.09***	5.58	2.55
Hispanic, no race	0.304	−0.106	−0.0763	---
Hispanic white	0.294	0.181	1.774	0.814
NonHispanic Asian	0.243	0.392	−0.414	1.283
NonHispanic black	0.105	0.268	2.410	1.373
NonHispanic white	−0.332	−0.163	1.301	0.930
Joint test chi2	6.56 (5 df)	9.34 (5 df)	10.79 (5 df)	2.98 (4 df)
Catholic	0.0973	−0.189	−0.0879	−0.459
Orthodox Christian	0.476	0.134	0.143	0.0227
Protestant	0.201	−0.283	−0.00645	−0.330
Muslim	0.444	−0.162	−0.212	0.305
Jewish	0.151	0.0654	−1.461	0.949
Buddhist	0.1599	0.253	−0.345	0.103
Hindu	−0.135	−0.336	0.731	−1.476
Other religion	−0.572	−0.911	0.345	−1.728
Childhood religion missing	−0.0806	−0.0592	1.071	0.172
Joint test chi2 (9 df)	5.87	9.18	5.44	9.50
Family income below average	−0.407	−0.0586	1.014	−0.864
Family income average	−0.217	0.225	.966	−0.802
Family income above average	−0.600	0.191	.379	−0.597
Family income far above average	−0.449	−0.0260	2.131	−1.801
Joint test chi2 (4 df)	8.17	6.84	4.421	3.40
Schooling	0.0216 (0.133)	−0.00103 (0.08)	−0.00977 (0.22)	−0.000549 (0.02)
Visa depression	−0.248 (1.61)	−0.192 (1.75)	−1.156 (2.56)**	−0.621 (1.61)
Spouse of FB citizen	−0.446	−0.121	NA	NA
Parent of U.S. citizen	−0.359	−0.404	NA	NA
Child of U.S. citizen	−0.492	−0.405	NA	NA
Adult unmarried child of U.S. citizen	−0.164	0.0646	NA	NA
Sibling of U.S. citizen	−0.156	0.819	0.129	−0.444
Spouse of LPR	−0.958	−0.292	NA	NA
Child of LPR	−0.422	−0.0214	NA	NA
Employment	2.380	2.298	−2.279	0.223
Diversity	2.820	2.287	−0.628	1.525
Refugee/asylee/parolee	0.327	0.358	−2.653	1.820
Legalization	−0.0856	−0.0902	---	---
Joint test chi2	157.01*** (12 df)	180.64*** (12 df)	10.89* (4 df)	20.24*** (4 df)
Adjustee, nonimm visa	0.309	0.0506	2.051	−0.584
Adjustee, EWI/WI	−0.342	−0.667	3.475	−0.277
Adjustee, UU/UN	−0.232	−0.549	---	0.434
Adjustee, no code	0.327	−0.0671	1.499	−0.325
Adjustee, Warren overstay	0.248	0.0391	2.779	0.929
Joint test chi2	12.43* (5 df)	20.04** (5 df)	10.01* (4 df)	4.61 (5 df)
Number of observations	3566	3577	333	537

Notes: NA = not applicable. Specifications also include origin-area fixed effects. Absolute values of asymptotic t-ratio appear in parentheses under parameter estimates for numeric and binary variables. Joint tests reported for multiple-category categorical variables.

^*p < .05,

^**p < .01,

^***p < .001

two-tailed tests for single coefficients, one-tailed for joint tests

Visa depression inhibits the me-principal assertion, among both men and women and both principals and nonprincipals, attaining statistical significance in the men nonprincipals equation. Thus, if declaring oneself the principal helps to repair the self, then immigrants who have endured visa depression cannot help themselves even in this way. It would seem that visa depression reduces the drive to assert oneself the principal, even among principals.

The two age variables are jointly highly statistically significant in both the sex-specific principals equations and borderline in the men's nonprincipals equation. In these three equations, the parabolas open downward, with peaks at 57 and 30 years among male principals and nonprincipals, respectively, and 79 among female principals. Thus, if, among principals, advancing age attenuates false modesty or increases knowledge, then women are on an unambiguously upward trajectory, while men hit a bound in their late fifties. Among nonprincipals, however, appropriating principal status is a young man's game, peaking early – before age thirty – and diminishing thereafter.

Visa class is highly statistically significant in all equations. Among principals of both sexes, the top three categories correctly identifying their principal status are employment and diversity principals and spouses of native-born citizens. These three categories are also associated with some of the highest levels of schooling and English fluency (Table 4). Spouses of U.S. citizens and employment principals must be sufficiently fluent in English to have attracted a sponsor; and, while many employment and all diversity principals have a schooling requirement (albeit one that can be waived given occupational credentials), spouses of native-born U.S. citizens reflect Americans' penchant for assortative mating in schooling (Jasso, Massey, Rosenzweig, and Smith 2000).

Meanwhile, the visa classes associated with the lowest probability of correctly identifying principal status are spouses of LPRs, children (age 18–21) of U.S. citizens, husbands of foreign-born U.S. citizens, and mothers of U.S. citizens. Whether these cases reflect lack of information or insufficient understanding of the term “principal” and its translation into the 94 non-English languages used in the New Immigrant Survey, one can only speculate. But it is interesting that they are all family immigrants, that most of the sponsors are foreign-born, and that while some are youthful (those with the child of U.S. citizen visa), others may be quite old (mothers of U.S. citizens).

Of course, the false modesty and self-effacement mechanism may also be operating; the spouses are all spouses of foreign-born persons, and the mothers and children are for the most part mothers and children of foreign-born persons. These results strongly complement the gender result in the pooled equations, and introduce an element of what may be called the immigration status hierarchy, with low-status persons -- dependent for their new visa on the higher-status naturalized U.S. citizens and previous immigrants -- displaying a reluctance to express their principalhood and making obeisance, as it were, to their status superiors.

The adjustee and adjusting-from-illegality variables are jointly statistically significant in both the sex-specific principals' equations and both the men's equations, thus resembling the age variables. Among principals, the two lowest probabilities of asserting principal status are among the formerly illegal EWI/WI immigrants and the set with a UU/UN code. In contrast, among male nonprincipals, the EWI immigrants are most likely to say they are principals. At first blush, these results appear to be at odds with each other. However, both responses have a powerful element in common. The EWI immigrants have lived for years in the shadows, and as a survival strategy have learned to dissemble and to conceal. Here we find principals reluctant to reveal their principal status and nonprincipals appropriating it. These may be vestigial behaviors, as the formerly illegal emerge from the shadows and start to shed the habits of illegality.

Childhood religion does not reach statistical significance. However, it is interesting that the highest probability of the me-principal assertion among women nonprincipals is found among Jewish and Muslim women; the lowest probabilities are found among Hindu and other-religion women. The Muslim effect is provocative, as it may signal a mechanism for coping with gender inequality.

The coefficients for parental relative family income, though not statistically significant, hint at an interesting pattern. Among principals of both sexes, those from average socioeconomic backgrounds have the highest probability of correctly identifying that they are principals. Among male nonprincipals, the richest are most likely to erroneously claim that they are the principal.

Finally, schooling does not reach significance in any equation. We also tested for a direct effect of having documents lost, but did not find any. Lost documents, of course, operate through visa depression. The origin area fixed effects are statistically significant in three of the four sex-specific equations and borderline (prob value of .0521) in the men's nonprincipals equation.

More broadly, the gender difference in making the me-principal assertion raises the question whether a similar mechanism may be operating in social surveys – men systematically overstating, women systematically understating, their schooling and earnings -- contributing to the observed gender gap. Either the male or female component of such a mechanism would have far-reaching consequences, as discussed by Ruel and Hauser (2007). It is not often the case that survey data permit comparison of a respondent's real and reported characteristics. The NIS may thus be useful in assessing the broader conjecture.

6. RACE-ETHNIC COMPOSITION OF NEW IMMIGRANTS AND THE NEW BLACK IMMIGRATION FROM AFRICA

6.1. Immigration and the American Race/Ethnic Structure

A perennial theme in immigration research and policy involves the effects of immigration on the racial and ethnic composition of the United States (Alba and Nee 2003; Bean and Stevens 2003; Jasso and Rosenzweig 1990, 2006; Smith and Edmonston 1997). Indeed, the history of U.S. immigration law can be written from a race-ethnic perspective, with critical junctures – such as the first quantitative restrictions in 1921 and, later, the removal of quotas and racial bars --intimately tied to visions of the ideal racial and ethnic composition of the country.

In the 1970s, as it was becoming clear that the family reunification provisions of the 1965 Immigration Act engendered increased flows of relatives of previous immigrants, a new concern arose in policymaking circles. For persons in countries without a foothold in the immigration stream, there would be little possibility of immigrating to the United States. For example, documents of the U.S. Select Commission on Immigration and Refugee Policy, whose final report was issued in 1981, convey a sense of urgency about opening a new channel for “independent” immigration, and the oral tradition suggests that at least part of the concern involved the small numbers of black immigrants from Africa. According to the Staff Report of the Select Commission (1981:455–456),

[I]ndependent immigration would be used by nationals of many African and European countries, as well as by those of some of the currently more prominent countries of immigration. This new channel might, therefore, be expected to open immigration to new or renewed source of immigrants, while both it and the family reunification category would continue to build on the more recent bases of immigration.

A number of procedures for selecting immigrants in the envisioned open immigration channel were discussed, including an ill-fated point system (Jasso 1988). Eventually, however, the United States established the Diversity Visa Program, making available new visas for blacks and others from Africa. Note that there was no scarcity of black immigrants from the Caribbean; the dearth was of black immigrants from Africa.

However, assessing the race-ethnic composition of cohorts of new legal immigrants – and the success of the diversity lottery program – has not been possible given that the U.S. stopped collecting data on the race and ethnicity of new immigrants in 1961. The New Immigrant Survey, as noted earlier, includes the standard two questions on race and Hispanic origin, and thus enables for the first time since 1961 description of the race-ethnic composition of an immigrant cohort.³⁰

We focus first on the race-ethnic composition of the cohort and next on black immigrants.

6.2. Special NIS Data for Race-Ethnic Analysis

The two standard survey questions on race and Hispanic origin described earlier will receive special attention in this section because of uncertainty surrounding the approximately 7.6% of respondents who did not answer the race question and who may affect the estimated proportion black. A third piece of data is introduced here, both for its own intrinsic interest as well as to deepen understanding of nonresponse to the race question – the respondent's skin color as coded by the interviewer.

Skin color appears prominently in many discussions of immigration, such as Alba and Nee (2003), Bean and Stevens (2003), and Jensen, Cohen, Toribio, DeJong, and Rodriguez (2006), as well as in more general discussions of the American stratification structure, for example, Anderson (1999), Feagin (1991), Gans (2005), Lacy (2007), and Massey and Denton (1992). The New Immigrant Survey measured respondent skin color on an 11-point scale, ranging from zero to 10, with zero representing albinism (the total absence of color) and 10 representing the darkest possible skin. The ten shades of skin color corresponding to the points 1 to 10 on the NIS Skin Color Scale are depicted in a chart, with each point represented by a hand, of identical form, but differing in color. The NIS Skin Color Scale is for use by interviewers, who “memorize” the scale, so that respondents never see the chart. [A copy of the Scale is reproduced in the introduction to this special issue.]

Skin color assessments are available for 4,652 main sampled immigrants (54.3%) in the Adult Sample, as follows. First, skin color was not assessed among the 321 overseas respondents (section 2.4 and Appendix Table A.1), although they were asked the Hispanic origin and race questions. Second, skin color was assessed among non-overseas respondents who were interviewed in person. Third, skin color was assessed among non-overseas respondents who had met the interviewer even if the interview was conducted (or completed) by telephone. The data indicate that 29% (27% weighted) of the skin color ratings were made for respondents interviewed in person. Respondents interviewed by telephone tend to fall into two main groups, those who requested an interview language in which a fluent interviewer or interpreter was not available on site, and those who requested a telephone interview (for their convenience or for privacy, etc.).

Interviewers may have systematically seen respondents of differing skin color (given the link between skin color and interview language), or, alternatively, they may perceive skin color differently. Thus, although the data provide interviewer ID codes and it is straightforward to control for threshold effects by incorporating interviewer fixed effects in estimation, it is not clear whether the color scale should be corrected for interviewer effects. Further work on this question is warranted, including calibration and physical measurements in future rounds of the NIS. Moreover, analysts can test for the possibility that skin color perception differs with duration of exposure by including a binary variable for the mode of the interview.

An additional point is worth noting. A few months after the start of fieldwork NIS survey managers became concerned that interviewers were using too many “zeroes,” which were to be reserved for albinism, and issued a memorandum to all staff on this matter; accordingly, analysts can test for effects of the memorandum by including a binary variable for whether the interview took place before or after the date of the memorandum.

In the work reported below, we distinguish between the “raw” interviewer ratings and “corrected” ratings obtained by regressing the raw rating on the date and mode binary variables and the interviewer fixed effects.

In this section we also use the behavioral measure of English fluency shown in Table 4. This binary measure is coded “1” if the respondent chose to be interviewed in English and, further, did not use any other language during the interview. This is a stringent measure, one that indicates sufficient fluency in English to sustain a long and elaborate interview.³¹

We also introduce childhood language and origin-country official language in this section. The New Immigrant Survey asked the main sampled adult respondents (except the 321 overseas respondents) the question, “What languages did you speak in your home with your parents when you were age 10?” We coded the responses into six categories. The categories and their proportions among the main sampled immigrants are: English only, 6.99%; Spanish only, 35%; English and Spanish, 1.34%; English and a language other than Spanish, 5.46%; Spanish and a language other than English, 1.01%; English, Spanish, and another language, .20%; and other, 49.7%. Data are missing for .32%.

We constructed a binary variable indicating whether English is an official or dominant language of the country of birth. Approximately 25.4% of the sample come from a country where English is an official or dominant language.

6.3. Race, Hispanic Origin, and Skin Color in the 2003 Immigrant Cohort

Table 15 reports the race-ethnic composition for the immigrant cohort, together with the corresponding figures for the U.S. population in 2003. The racial composition of the two populations differs in two main ways: First, the proportion white among new immigrants is dramatically lower than in the U.S. population as a whole – 48% versus 81%. Second, the proportion Asian among new immigrants is, again, dramatically higher than in the total U.S. population – 29% versus 4.1%. The proportion black in the immigrant cohort is 11.2%, or about 1.5% less than the proportion black in the resident population, but, given that a nontrivial proportion of new immigrants did not report their race − 7.6% – the true proportion black in the immigrant cohort may be the same or even higher than among residents.³²

Table 15.

Percentage Distribution of Race and Hispanic Origin: U.S. Resident Population in 2003 and NIS-2003 Cohort

Race and Hispanic Origin	U.S. Resident Population in 2003	NIS-2003
		Race	Race-Hispanic
White	80.5	48.1
White, Hispanic			28.5
White, not Hispanic			19.5
White, no Hispanic information			.12
Black	12.7	11.2
Black, Hispanic			.56
Black, not Hispanic			10.6
Black, no Hispanic information			.04
American Indian	.959	2.51
American Indian, Hispanic			.46
American Indian, not Hispanic			2.05
Asian	4.10	28.6
Asian, Hispanic			.41
Asian, not Hispanic			28.2
Asian, no Hispanic information			.05
Pacific	.171	.77
Pacific, Hispanic			.18
Pacific, not Hispanic			.59
Two or more races	1.48	1.15
Mixed-race, Hispanic			.83
Mixed-race, not Hispanic			.32
No race	---	7.60
No race, Hispanic			5.63
No race, not Hispanic			1.61
No race, no Hispanic information			.36
Hispanic origin	13.7		38.1
Not Hispanic origin	86.3		61.3
No information on Hispanic origin	---

With respect to Hispanic origin, the figures in Table 15 are no surprise. The proportion of new immigrants reporting Hispanic origin is almost 3 times as large as among the total population – 38.1% versus 13.7%. Among respondents reporting themselves as Hispanic, 75% report themselves as white – substantially more than the 50% in the general population (Alba and Nee 2003:9). Further, and more to our purpose in this section, 14.8% of the self-reported Hispanics did not report their race.

Nonresponders to the NIS race question included a large majority who reported themselves as Hispanic (74%). A closer look at the source data indicate that the top four origin countries of race nonresponders, which together comprise two-thirds of the nonresponders --Mexico (40.3%), El Salvador (13.1%), Dominican Republic (7.92%), and Guatemala (5.28%) – are countries whose immigrants overwhelmingly report themselves as Hispanic (over 97% in every case). Within these four countries, the proportion who did not respond to the race question hovered around 16–17% in three of them (Mexico, El Salvador, and Guatemala), but registered a larger 26.5% for the Dominican Republic, suggesting that these two subsets may differ in pertinent ways. A natural question is whether the nonresponders resemble the responders. Among those who did report a race, while the proportion black did not reach even half of one percent for Mexico, El Salvador, and Guatemala, for the Dominican Republic the figures are 6.85% black alone and 12.3% black and another race, for a total of 19.1 percent. Thus, if nonresponders resemble responders, then some of the nonresponders in the Dominican Republic are black, thereby increasing the overall proportion black.

We next examine the interviewer-coded skin color by nonresponse on the race question. In all four countries, average skin color is darker among nonresponders than among responders; however, the magnitude of the discrepancy is small in all cases except that of the Dominican Republic, where the difference is over one unit on the 11-point scale.

These results suggest, first, that the proportion black among the new immigrants may indeed be higher than our estimate (11.2%) – and perhaps even higher than the proportion black in the U.S. population (12.7%), and, second, that among black immigrants, the relative size of the subset born in the Americas may be larger than our estimate below (47.2%).

These results also suggest that some immigrants who see themselves as Hispanic are reluctant to assign themselves one (or more) of the five races – though at 15% substantially fewer than the 40% who chose the “some other race” option provided in the 2000 Census (Tienda and Mitchell 2006) – and that, consistent with Waters (1990, 1999), Hispanic immigrants from the Dominican Republic are especially reluctant to do so, as are those of darker skin color. As Tienda and Mitchell (2006:44) suggest, some Hispanics may not find any of the five official races a good fit.

It is also interesting that few respondents from Latin America choose both white and American Indian races, notwithstanding the history and explicit ideology of mestizaje (Villarreal 2010). On the other hand, the question wording may inhibit selection of the American Indian response option: “A person having origins in any of the original peoples of North, Central, or South America, and who maintains tribal affiliation or community attachment.” The first clause would be satisfied by much larger numbers than the second clause. It is interesting to speculate how changes in the wording of the race question – to highlight ancestry rather than identity or political affiliation – would affect response.³³

Table 16 reports average skin color for the race and Hispanic origin groups. Raw figures are based directly on the interviewer assessments. Corrected figures are based on regressions including the interviewer fixed-effects and the date and interview mode dummies. The raw and corrected estimates differ not only in magnitude but also in the relative orderings, although in both estimates the white group is the lightest and the black group the darkest. As already noted, average skin color is darker for groups that did not report a race – contrast the three subsets which did not report race (bottom rows in the top panel) with the entire relevant groups (bottom panel). For example, average skin color among all persons who reported that they are not of Hispanic origin is lighter than among the subset who reported that they are not of Hispanic origin but did not report a race (4.11 versus 4.49 in the raw score and 4.21 versus 4.53 in the corrected score). The effect of skin color on the propensity to assign oneself one (or more) of the five races clearly warrants further research.

Table 16.

Average Skin Color, as Reported by Interviewer, Raw and Corrected for Interviewer, Timing, and Mode Effects: NIS-2003 Cohort

Race and Hispanic Origin	Race		Race-Hispanic
	Raw	Corrected	Raw	Corrected
White	3.56	2.92
White, Hispanic			4.09	3.39
White, not Hispanic			2.56	2.44
White, no Hispanic information			---	---
Black	7.13	7.84
Black, Hispanic			7.68	7.58
Black, not Hispanic			7.10	7.80
Black, no Hispanic information			---	---
American Indian	4.56	4.20
American Indian, Hispanic			4.58	4.29
American Indian, not Hispanic			4.46	4.70
Asian	3.92	4.21
Asian, Hispanic			---	---
Asian, not Hispanic			3.92	4.15
Asian, no Hispanic information			---	---
Pacific	4.97	4.53
Pacific, Hispanic			---	---
Pacific, not Hispanic			4.71	4.43
Two or more races	4.27	4.67
Mixed-race, Hispanic			5.01	4.90
Mixed-race, not Hispanic			---	---
No race	4.78	4.47
No race, Hispanic			4.80	4.55
No race, not Hispanic			4.49	4.53
No race, no Hispanic info.			---	---

	Raw	Corrected
Hispanic origin	4.27	3.61
Not Hispanic origin	4.11	4.21
No information on Hispanic origin	5.17	4.82

Notes: Skin color is measured by the interviewer on an 11-point scale. Corrected skin-color averages are obtained from regressions that include, besides the race-Hispanic categories, interviewer fixed-effects and binary variables for whether the interview took place after the date of a memo to interviewers concerning the coding of skin color and for whether the interview was conducted in person. Average skin color not shown for cells with fewer than 20 cases.

6.4. Black Immigrants

An important theme in recent American history is the increasing diversity within the black population. At first the critical dimension of diversity appeared to be foreign birth, with most foreign-born blacks originating in Caribbean countries such as Jamaica and Haiti. Thus, there seemed to be “two kinds” of blacks in the United States, those whose ancestors had been forcibly brought to the United States as slaves and those who freely immigrated. But soon another dimension emerged -- the origin continent of black immigrants.

Though the history of the Diversity Visa Program is yet to be written, a lively oral tradition suggests that at least some of the impetus for the diversity lottery program came from policymakers holding dear the vision of an American people drawn from every corner of the globe and noticing the dearth of immigrants from Africa. Whatever its roots, there is little doubt that the Diversity Visa Program has substantially increased the flow of immigrants from Africa. In the period 2001–2009, the proportion Africa-born of all diversity-visa admissions to legal permanent residence ranged from 35.2% in 2005 to 50.4% in 2009 (U.S. INS and U.S. DHS, Yearbook of Immigration Statistics, 2001–2009).³⁴

The growing view is that there are “three kinds” of blacks in the United States today: (1) descendants of slaves (who also, importantly, endured the Jim Crow era – 1876–1965); (2) immigrants from the Americas and their descendants; and (3) voluntary immigrants from Africa and their descendants. Scholarly interest is increasing in understanding heterogeneity among blacks in the United States (Corra 2005; Elo, Mehta, and Huang 2011; Massey, Mooney, Torres, and Charles 2007; Portes and Rumbaut 1990).

A priori one would expect that new immigrants from Africa would be more highly self-selected and thus more highly skilled than new immigrants from the Americas. There are at least three mechanisms driving this conjecture, the first two related to the costs of migration, the third to U.S. visa allocation policy (described in section 2 above). The first mechanism pertains to the fact that Africa is more distant, and thus the costs of migration are higher. The second pertains to the fact that the African flow is a flow of pioneer immigrants, for whom the costs are higher than for Caribbean immigrants whose co-nationals have already established a beachhead and with whom they are already embedded in networks. The third mechanism highlights visa requirements – diversity visas and (most of the) visas for skilled immigrants have a schooling requirement, so that pioneer immigrants are likely to be more highly schooled than family reunification immigrants. The available data support this conjecture: Black foreign-born from Africa are indeed more skilled and earn more than black foreign-born from the Americas (Massey et al. 2007; Portes and Rumbaut 1990).

The New Immigrant Survey can shed light on the new black immigration, and in this section we provide the first quantitative description of a representative sample of a recent cohort of new legal immigrants. Future work should undertake systematic comparisons with the native-born, a task beyond the scope of the present paper.

Among the main sampled adult immigrants, 11.6% reported that they are black, inclusive of mixed-race origin (a total of 1,107 immigrants). Within this set, those reporting nonHispanic origin constitute 10.8% of the sample, while blacks of Hispanic origin are about .77% of the sample. Over half of the black immigrants -- 51.4% -- come from African countries, 47.2% from the Americas, and 1.45% from the rest of the world.³⁵

Among the 1,107 black immigrants, 1,079 cases reported only a black race and 28 reported two or more races. Because the mixed-race immigrants are too few for reliable analysis, we eliminate them from the analyses in this section. Among the black-only immigrants, 52.7% are from Africa, 45.8% from the Americas, and 1.5% from the rest of the world.

The black-only immigrants from outside Africa and the Americas number only 14, and thus, for the rest of this section, analysis is restricted to the Africa and Americas immigrants. The final sample of immigrants who reported being black only and who come from Africa or the Western Hemisphere totals 1,065. In this final black sample, 53.5% are from Africa and 46.5% from the Americas.

The two top African origin countries are Nigeria (13%) and Ethiopia (11.3%), and the two top Western Hemisphere origin countries are Haiti (17.8%) and Jamaica (14.5%).

Table 17 reports the main characteristics of the black-only immigrants, separately by origin area. As expected, the two immigration streams differ in important ways. The Americas stream is more established, and hence the proportion of immigrants with family visas is substantially larger - approximately 56.5% versus 15.9%, not counting spouses of U.S. citizens, and 87.6% versus 40.4%, including spouses. Indeed, the Americas stream is so well-established and robust that the two top countries, Haiti and Jamaica, are not even eligible for diversity visas, and neither is the Dominican Republic; the proportion of Americas-born blacks with lottery visas is a negligible tenth of one percent. In contrast, 40% of the Africa-born blacks achieved LPR with diversity visas, consistent with one of the original driving ideas behind the lottery program. Moreover, the proportion with employment visas is negligible in both streams (1% among the Africa-born and 1.7% among the Americas-born). Finally, the proportion with refugee/asylee/parolee visas is almost twice as large among the Africa-born, reflecting their origin-country experiences - over three times as many Africa-born as Americas-born suffered harm before coming to the United States.

Table 17.

Characteristics of Black Immigrants from Africa and the Americas: NIS-2003 Cohort

Characteristic	Born in Africa	Born in the Americas
Percent female	50.4	54.1
Average age at admission to LPR (years)	35.7	39.3
Spouse of native-born U.S. citizen (%)	12.5	10.8
Spouse of foreign-born U.S. citizen (%)	12.0	20.3
Parent of U.S. citizen (%)	8.53	12.7
Minor child of U.S. citizen (%)	3.30	9.31
Employment principal (%)	1.01	1.72
Diversity principal (%)	29.8	0.11
Spouse of diversity principal (%)	10.0	.34
Refugee/asylee/parolee principal (%)	14.0	7.25
Other family visa (%)	4.09	34.5
Other visa (%)	4.91	3.02
Adjustee (%)	43.4	48.3
Documents lost during visa process (%)	11.3	12.3
Duration of visa process (years)	3.13	5.33
English an official language (%)	59.7	53.0
Spoke English only at age 10 (%)	4.82	47.1
Spoke English (only/some) at age 10 (%)	24.0	51.8
Suffered harm outside the U.S. (%)	14.0	3.51
Average schooling (years)	13.1	11.2
Own home (%)	16.9	18.1
Own home, new arrivals (%)	4.23	7.91
Own home, adjustees (%)	32.8	29.0
Intend to stay in United States (%)	77.9	71.4
Interviewed in English only (%)	75.9	65.0
Average skin color (raw)	7.46	6.81
Average skin color (corrected)	7.37	7.10
Percent in black-only sample	53.5	46.5
Sample size	688	377

Notes: Sample restricted to immigrants who reported only black race. Lost documents calculated on principals only.

Table 18 reports previous illegal experience in the black sample, paralleling the estimates for the whole sample in Table 6. The proportion with previous illegal experience is larger among the Americas-born than among the Africa-born (5.18% versus 3.04% in the lower-bound estimate and 40.9% versus 15.7% in the largest estimate obtained when all components from the administrative record as well as the survey measures are included (section 2).

Table 18.

Previous Illegal Experience, by Information Source: NIS-2003 Cohort

Source of Information	Born in Africa	Born in the Americas
Immigrant legalization visa	1.18	.66
Nonimmigrant code EWI/WI	1.86	4.52
Nonimmigrant code UU/UN	.60	1.61
Nonimmigrant code missing	7.79	13.3
Nonimmigrant Warren measure	2.11	16.4
Survey measures	4.16	13.4
Total based on leg visa or EWI code	3.04	5.18
Total based on record alone	12.4	35.8
Total including survey measures	15.7	40.9

Notes: The information based on immigrant and nonimmigrant visa is from the official administrative immigrant record. The four components based on nonimmigrant visas are mutually exclusive. The Warren measure refers to having a nonimmigrant tourist visa (B2) and reporting the most recent entry six years or more earlier. The two other components – the immigrant legalization visa and the survey measures – may be combined with any of the other codes, so that, for example, a given respondent may be included in both the legalization visa figure, one of the nonimmigrant visa figures, and the survey measures. Thus, the total estimate is less than the sum of the components. The total estimate may be an underestimate, as it may miss new-arrival immigrants who were already living in the United States illegally and do not report it in the survey measures, as well as persons who had been working without authorization.

Almost every characteristic listed in Tables 17 and 18 merits sustained scrutiny. In this paper, however, we focus on two important indicators of skill and of potential for social and economic incorporation (Alba and Nee 2003; Bean and Stevens 2003; Jasso and Rosenzweig 1990, 2006; Portes and Rumbaut 2006) -- schooling and English fluency, the latter manifested in choosing to be interviewed in English and completing the entire interview exclusively in English. The Africa-born immigrants have completed on average two more years of schooling than the Americas-born, even though the Americas-born are older by almost four years, on average. The Africa-born were interviewed in English at a rate over ten percentage points larger than the Americas-born, mirroring the larger percentage born in a country where English is an official or dominant language, although a substantially larger fraction of the Americas-born spoke English as a child (over nine times as many Americas-born blacks as Africa-born blacks spoke English only at age 10, and about twice as many spoke at least some English at age 10). Below we test for differences in schooling and English fluency between the Africa-born and the Americas-born in multivariate models.

Finally, both the raw and corrected skin color scores indicate that the Africa-born are darker, on average, than the Americas-born, although the magnitude of the differential does not reach even one unit on the skin-color scale (.65 on the raw scale and .27 on the corrected scale). If, across other immigrant cohorts, the Africa-born blacks indeed are darker - and also more accomplished than the Americas-born blacks -- then the usual skin color correlations would be overturned. Such patterns could be monumental for many aspects of the future of the United States. They could lay siege to stereotypes linking skin color to educational attainment and productivity, possibly even hastening racial integration.

6.5. The Determinants of Schooling and English Fluency among Black Immigrants

Table 19 reports OLS estimates of the schooling equation and binary logit estimates of the English fluency equation. In the pooled equations (not shown), women have less schooling than men (with statistical significance beyond .003) and they are less likely to be fluent in English (with statistical significance beyond .023). The big story, however, pertains to the different streams of black immigrants from Africa and from the Americas.

Table 19.

Determinants of Schooling (OLS Estimates) and English Fluency (Logit Estimates), Black Immigrants from Africa and the Americas: NIS-2003 Cohort

Regressor	Years of Schooling		Fluency in English
	Men	Women	Men	Women
Age	0.509	.310	.0450	.00738
Age squared	−0.00625	−.00428	−.000848	−.000263
Joint test F	27.6*** (2, 547)	14.3*** (2, 481)	NA	NA
Joint test chi2 (2 df)	NA	NA	3.74 (2 df).	.93 (2 df)
Spouse of NB citizen	.395	1.411	−.00385	−.540
Spouse of FB citizen	.599	1.190	−2.301	−.516
Parent of U.S. citizen	.0205	−.712	−1.441	−1.978
Child of U.S. citizen	2.128	2.098	−1.583	−1.071
Employment principal	1.751	2.750	.189	---
Diversity principal	.825	2.378	−1.393	−1.051
Spouse of diversity prin.	1.676	.556	−1.154	−1.248
Refugee principal	−.914	−2.272	−1.964	−.938
Other family visa	.0739	2.510	−1.180	−1.249
Joint test F	2.87** (9, 547)	5.23*** (9, 481)	NA	NA
Joint test chi2	NA	NA	9.35 (9 df)	7.29 (8 df)
Adjustee	1.233 (2.26)*	1.453 (2.22)*	2.202 (4.13)***	1.030 (1.68)
Previous illegal experience	−.181 (.44)	.390 (.67)	−.431 (.72)	.151 (.29)
Born in Africa	2.406 (4.53)***	1.346 (2.33)*	1.005 (2.06)*	1.019 (2.49)*
English an official language	NA	NA	3.276 (7.33)***	3.340 (8.00)***
Spoke English only at age 10	NA	NA	2.082	2.394
Spoke some English at age 10	NA	NA	.884	1.125
Joint test chi2 (2 df)	NA	NA	14.6*** (2 df)	25.5*** (2 df)
Constant	1.350 (.73)	4.532 (2.20)*	−.640 (.32)	−.758 (.56)
R-squared	.298	.330	NA	NA
Log pseudolikelihood	NA	NA	−199.76	−173.10
Observations	562	496	563	501

Notes: NA = not applicable. Absolute values of asymptotic t-ratio appear in parentheses under parameter estimates for numeric and binary variables. Joint tests reported for multiple-category categorical variables.

^*p < .05,

^**p < .01,

^***p < .001;

two-tailed tests for single coefficients, one-tailed for joint tests

Table 19 confirms that, net of age, and the visa, adjustee, and previous illegal experience variables, black immigrants born in Africa have statistically significantly higher educational attainment than those born in the Americas - on average, 2.4 years more among men and 1.3 years more among women. Similarly, the Africa-born have a statistically significantly higher probability of being fluent in English than the Americas-born. This is net of coming from a country in which English is an official language and net of knowledge of English in childhood, which have their own substantial and highly statistically significant effects.

Thus, the Africa-born immigrants appear to be more intensely positively self-selected than the Americas-born, consistent with the recency of the immigration streams and the longer distance to the United States.

6.6. Black Immigration and the Attack on American Apartheid

The infusion of highly accomplished black immigrants threatens the very foundation of the racial stereotypes and racial hierarchies christened “American Apartheid” by Massey and Denton (1992). Such flows of highly accomplished black immigrants would erode “whites' assumptions about the social meanings attached to skin color” (Alba and Nee 2003:291). Among new immigrants, it is clearly the case that black immigrants from Africa have higher schooling and black immigrants from both Africa and the Americas have higher English fluency than the rest of the cohort. Average schooling completed is 11.9 years in the rest of the cohort, relative to 13.1 among the Africa-born and 11.2 among the Americas-born (Table 17). The proportion interviewed exclusively in English is 37.5% among nonblacks, versus 75.9% among Africa-born blacks and 65% among Americas-born blacks (Table 17). If black immigrants are compared to nonHispanic white immigrants, they lose top rank in schooling (nonHispanic whites average 14.1 years of schooling) but retain it in English fluency (57.2% of nonHispanic whites were interviewed exclusively in English).

Thus, racial differences are attenuated among immigrants, and infusions of immigrants may, depending on relative numbers and schooling averages, overturn the racial hierarchy in the United States.

Meanwhile, it is illuminating to also consider the association between skin color and the schooling and language outcomes. Black immigrants are darker than all other immigrants (Table 16), and thus it can be said that among immigrants, the darkest ones are among the most accomplished. What about the contrast between the two streams of immigrants? The evidence is not conclusive. As shown in Table 17, average skin color is indeed darker among the Africa-born; however, the discrepancy is small in the corrected measure. If future rounds of the NIS obtain physical measurements of skin color or use calibration procedures, it will be possible to definitively contrast the skin color of the Africa-born and the Americas-born. If even within the set of black immigrants, the darker are the more accomplished, the foundation for a stratification system based on skin color would be shattered.³⁶

It is widely thought -- building on Goffman's (1963) analysis of stigma -- that an entrenched and persistent “black stigma” (Feagin 1991) or “racial stigma” (Loury 2003) operates to eclipse the accomplishments of U.S. blacks, leaving visible only the negative stereotypes and images of a less productive minority of the U.S. black population. Note that black immigrants, especially those from Africa, could join accomplished native U.S. blacks to achieve a surpassing critical mass that would, in Gans' (1999:381) evocative words, “disturb white America's long association of poverty with blackness,” generating new and positive stereotypes and images and bringing to life the brightest of Gans' (1999:381) scenarios for the future of race in America.

Interestingly, and perhaps ironically, it is immigration, which has sometimes been blamed for the lack of more rapid progress among American blacks, that may prove to be the avenging angel that obliterates the color line - which dominated the twentieth century, as predicted by DuBois (1903) - and catapults the American black population onto the overworld. Indeed, in such a future there would no longer be white Americans or black Americans but only Americans with a variety of ancestral histories.

Finally, it is illuminating to consider that the current President of the United States exemplifies the variety of ancestral histories – with a white American mother and a black African father – and has been joyously embraced by both white and black Americans. Further, with every passing generation, the within-individual variety of ancestral histories increases.

7. SKIN COLOR AND SPOUSE SELECTION AMONG U.S. CITIZEN SPONSORS OF IMMIGRANT SPOUSES

Two icons with roots in the American Midwest illuminate this discussion – Marion Robert Morrison (1907–1979), better known as John Wayne, and Stanley Ann Dunham (1942–1995), better known as the mother of the 44^th President of the United States, but an accomplished anthropologist in her own right. Between them they had three wives – one born in the United States to parents from Spain, the others born in Mexico and Peru – and two husbands – born in Kenya and Indonesia.³⁷

7.1. Spouse Choice by Sponsor Gender and Nativity

As we have seen and as discussed in the literature (Alba and Nee 2003; Bean and Stevens 2003; Jasso and Rosenzweig 1990), immigration is increasing diversity in the U.S. population – by introducing substantial contingents of Asian and Hispanic immigrants. Moreover, as discussed above, immigration threatens to overturn racial and skin color associations with skill – by introducing accomplished black immigrants from Africa. We turn now to assess skin color and diversity in a more intimate arena, namely, within the married couples formed by U.S. citizen sponsors and their immigrant spouses. Note the crucial distinction between population diversity and within-couple diversity. Within-couple diversity would reflect the diminishing importance of skin color in spouse selection and thus in the broader society as well (Qian and Lichter 2007).³⁸

As noted in section 2, approximately a third of adults granted legal permanent residence every year are admitted as the spouses of U.S. citizens. In the NIS-2003, that figure is 34.1% (1,427 immigrants); approximately 16.2% are sponsored by native-born U.S. citizens and 17.9% by previous immigrants who have naturalized (Table 4). Within the set of spouses of U.S. citizens in the NIS-2003, a majority are sponsored by naturalized citizens (52.6% versus 47.4%), and, similarly, a majority are sponsored by men (62.9% versus 37.1%). The breakdown by sponsor gender-nativity is: NB men, 28.3%; NB women, 19.2%; FB men, 34.7%; and FB women, 17.9%.³⁹

By far the favorite place where U.S. citizens find mates is Mexico -- 24.1% and 24% in the sets sponsored by native-born and foreign-born, respectively. The second-place countries are the Philippines for native-born U.S. citizens (4.62%) and India for foreign-born U.S. citizens (6.05%). Patterns of spouse selection tend to differ for the naturalized and the native-born who sponsor spouses, and thus only two countries are in the top five in both sets (Mexico and the Philippines). For example, countries with a substantial U.S. military presence tend to provide spouses for native-born U.S. citizens (Jasso and Rosenzweig 1989:876–880, 1990:166–171; Bean and Stevens 2003:197–198).

A majority of the Mexico-born contingent of spouses are women sponsored by U.S. citizen men (55.3%). While there are roughly equal proportions of Mexico-born immigrant men sponsored by native-born and foreign-born U.S. citizens – 22.7% and 21.9%, respectively – the nativity breakdown differs among the U.S. citizen male sponsors of Mexico-born immigrant women – 30.5% foreign-born versus 24.8% native-born.

The question thus arises whether there is a skin color difference between the sponsors and their immigrant spouses and whether such a difference varies by sponsor nativity and gender. In the analyses that follow we distinguish between four kinds of married couples, formed by eight kinds of persons. The four types of married couples are formed by gender and nativity: (1) native-born U.S. citizen women and their immigrant husbands; (2) native-born U.S. citizen men and their immigrant wives; (3) foreign-born U.S. citizen women and their immigrant husbands; and (4) foreign-born U.S. citizen men and their immigrant wives.

Here our focus is on skin-color diversity, but we note that future research on these four types of married couples may prove useful in assessing patterns of immigrant incorporation, building on the insights that marital sponsorship provides a superior immigrant screening mechanism because spouses screen for long-term economic success rather than for a specific job (Jasso and Rosenzweig 1995) and that intermarriage by nativity is “the litmus test of assimilation” (Alba and Nee 2003:90) and promotes “quicker social integration” (Bean and Stevens 2003:176).

7.2. Special NIS Data for Studying Marriage and Diversity

To examine skin color patterns in these couples we use the skin color scale introduced in Section 6 and we construct two new variables to measure marital skin color difference. The first measure, applicable to spouses sponsored by U.S. citizens and LPRs, is defined as the immigrant's skin color minus the sponsor's skin color. This measure equals zero if both spouses have identical scores on the skin color scale; it is positive if the immigrant is darker than the sponsor, and negative if the immigrant is lighter than the sponsor. The second measure, applicable to all married couples, is defined as the husband's skin color minus the wife's skin color. This measure equals zero if both spouses have identical scores on the skin color scale; it is positive if the husband is darker than the wife, and negative if the husband is lighter than the wife. The second measure enables direct examination of the longstanding insight that within all human groups or populations, men are slightly darker than women (van den Berghe and Frost 1986; Jablonski 2004; Jablonski and Chaplin 2000).

The NIS data include 536 spouse-of-U.S.-citizen couples with skin color scores for both spouses. The breakdown by sponsor gender-nativity is: NB men, 24.3%; NB women, 19.4%; FB men, 36.3%; and FB women, 19.9%. Within this set, approximately 96% (514 couples) were interviewed by the same interviewer. Thus, under the assumption that interviewer effects, if any, are mainly threshold effects, we use the raw scores as the underlying measure for the new skin-color difference variable.⁴⁰

7.3. Skin-Color Patterns Among Married Couples Formed by U.S. Citizen Sponsors and Their Immigrant Spouses

To begin, we examine the eight skin-color distributions among gender-nativity subsets of sponsors and their immigrant spouses. For each of the eight pairs formed by these eight distributions (such as the pair formed by native-born U.S. citizen sponsor women and foreign-born U.S. citizen sponsor women), as well as for each of the four pairs formed by the pooled distributions (such as the pair formed by all foreign-born sponsors and all native-born sponsors), we use the Kolmogorov-Smirnov test to test the hypothesis that both distributions in the pair are drawn from the same underlying distribution. Proceeding in this way, we establish that sponsor skin color differs by nativity but not by gender, while immigrant skin color differs by gender but not by sponsor nativity. These results (not shown) are strong and hold in all comparisons. For example, the result that sponsor skin color differs by nativity emerges in three tests: (1) between all native-born sponsors and all foreign-born sponsors (K-S = .249, p = .000); (2) between male native-born sponsors and male foreign-born sponsors (K-S = .285, p = .000); and (3) between female native-born sponsors and female foreign-born sponsors (K-S = .234, p = .003).

Table 20 reports the means and standard deviations of skin color in the eight personal skin-color distributions. As expected from the Kolmogorov-Smirnov tests, which also provide directional information, foreign-born sponsors are darker than native-born sponsors, by approximately one point on the skin color scale, on average (3.92 versus 2.93 among men and 3.95 versus 3.04 among women). However, within nativity subset, male and female sponsors have similar skin color, on average (2.93 and 3.04 among the native-born and 3.92 and 3.95 among the foreign-born).

Table 20.

Summary Characteristics of Spouses' Skin Color and Skin-Color Difference in Marriages Formed by U.S. Citizen Sponsors and Their Immigrant Spouses: NIS-2003 Cohort

Sponsor Characteristics	Husbands	Wives	Skin-Color Difference	N
Sponsor male native-born	2.93 (2.17)	3.58 (2.09)	.651 (1.91)	133
Sponsor male foreign-born	3.92 (1.93)	3.65 (1.90)	−.267 (1.39)	193
Sponsor female native-born	4.60 (2.24)	3.04 (2.08)	1.57 (2.34)	106
Sponsor female foreign-born	4.22 (2.36)	3.95 (2.36)	.269 (1.74)	104

Notes: Skin color is measured by the interviewer on an 11-point scale. Skin-color difference is defined as the immigrant's skin color minus the sponsor's skin color. Estimates based on weighted data. Standard deviations in parentheses beneath arithmetic means. The average skin-color difference across all four sets of spouse-of-U.S.-citizen couples is .419 and the standard deviation is 1.92.

Meanwhile, again as expected from the Kolmogorov-Smirnov tests, female immigrants are lighter than male immigrants, on average (3.58 versus 4.60 among those married to native-born sponsors and 3.65 versus 4.22 among those married to foreign-born sponsors). However, within gender subset, immigrants married to native-born and foreign-born sponsors have similar skin color, on average (4.60 and 4.22 among males and 3.58 and 3.65 among females).

Figure 4 provides visual representation of these patterns, presenting the quantile functions for the eight distributions, grouped, as in Table 20, by the sponsor's gender and nativity characteristics. The sponsor distributions are represented by a connected line, and the immigrant distributions by an unconnected line. For any percentage rank, the higher the observation the darker the person. As just discussed, the sponsor distributions are very similar within nativity (in panels A and B for native-born and in panels C and D for foreign-born), while the immigrant distributions are very similar within gender (in panels A and C for men and in panels B and D for women).

Figure 4.

Quantile Functions of Skin-Color Distributions in Marriages Formed by U.S. Citizen Sponsors and Their Immigrant Spouses: NIS-2003 Cohort. Unconnected lines represent the sponsors' distributions.

The more striking feature in Table 20 and Figure 4, however, pertains to skin-color difference and, in particular, the skin-color difference in couples with a native-born female sponsor (panel A). As reported in Table 20, the average skin-color difference is positive in three subsets of the couples (the exception being couples with a foreign-born male sponsor), but the magnitude of the skin-color difference is larger by far in the subset with a native-born female sponsor. Thus, native-born American women appear to be marrying spouses who are on average darker by a point-and-a-half on the skin-color scale.

For a closer look, we report in Table 21 the proportions of sponsors in each of the four married-couple sets who are marrying someone lighter, marrying someone of the same skin color, and marrying someone darker. As shown, the proportion marrying darker lines up exactly as the average skin-color difference, led by native-born U.S. citizen women (64.3%), followed by native-born U.S. citizen men (44.8%) and foreign-born U.S. citizen women (36.7%), and ending with foreign-born U.S. citizen men (15.6%). The proportion marrying lighter lines up in exactly the opposite way.

Table 21.

Proportions Marrying Lighter, Like, and Darker among U.S. Citizen Sponsors of Immigrant Spouses: NIS-2003 Cohort

Sponsor Characteristics	Percent Marrying Lighter	Percent Marrying Like	Percent Marrying Darker	N
Sponsor male native-born	14.8	40.4	44.8	133
Sponsor male foreign-born	29.2	55.2	15.6	193
Sponsor female native-born	13.7	22.1	64.3	106
Sponsor female foreign-born	23.8	39.6	36.7	104
All sponsors	21.6	42.0	36.4	536

Notes: Skin color is measured by the interviewer on an 11-point scale. Skin-color difference is defined as the immigrant's skin color minus the sponsor's skin color. Estimates based on weighted data.

Recall now that in groups around the world men are a little darker than women (van den Berghe and Frost 1986; Jablonski 2004; Jablonski and Chaplin 2000). Accordingly, it is no surprise that husbands are darker than wives in three of the four subsets. What is worthy of note is that the magnitude of the difference is larger when the sponsor is a native-born U.S. woman. Further, what is also worthy of note is that native-born U.S. citizen husbands are lighter than their wives. Thus, there is an unambiguous nativity effect in skin color difference between husbands and wives.

As an initial attempt to quantify the nativity effect, we take the difference-in-differences (DD). Among women sponsors, the DD is 1.3 (1.57 minus .269); among men sponsors, the DD is −.918 (.651 minus −.267). Thus, the net amount by which native-born women marry darker is 1.3 units on the skin color scale, and the net amount by which native-born men marry lighter is .92 on the scale.

From a stratification perspective, it would appear that native-born U.S. citizens, and especially women, are leading the charge against colorism in this most intimate of spheres. They are pushing the boundaries of skin-color difference within marriage.

7.4. Multivariate Results on Marriage and Skin-Color Diversity

To examine the robustness of the patterns just discussed, we estimate regressions of skin-color difference on sponsor characteristics, including age, race, and Hispanic origin, and on the immigrant's birth area, separately by gender, and also both in the full sample and the subsample in which both spouses had the same interviewer. As shown in Table 22, multivariate analysis supports the early results. Native-born U.S. citizen women marry darker than foreign-born U.S. citizen women, and native-born U.S. citizen men similarly marry darker than foreign-born U.S. citizen men.

Table 22.

Skin-Color Difference Between U.S. Citizen Sponsors and Their Immigrant Spouses: NIS-2003 Cohort

Regressor	Sample of Married Couples Formed by U.S. Citizen and Sponsored Spouse		Subsample in Which Both Spouses Had the Same Interviewer
	Male Sponsors	Female Sponsors	Male Sponsors	Female Sponsors
Sponsor's age, sex, nativity, and race-ethnic group
Age	−0.0284	0.281	−0.0359	0.242
Age squared	0.000365	−0.00369	0.000438	−0.00322
Joint test F	.25 (2, 296 df)	8.39*** (2, 180 df)	.29 (2, 281 df)	5.89** (2, 173 df)
Native-born female	NA	0.800 (2.97)**	NA	0.773 (2.90)**
Foreign-born female	NA	NA	NA	NA
Native-born male	0.504 (2.25)*	NA	0.447 (1.98)*	NA
Joint test F	NA	NA	NA	NA
Hispanic, no race	−0.0473	−0.962	−0.0353	−0.891
Hispanic white	−0.0288	−0.531	0.129	−0.282
NonHispanic Asian	−0.280	−0.262	−0.269	−0.387
NonHispanic black	−0.793	−1.169	−0.626	−0.927
NonHispanic white	1.170	1.152	1.183	1.257
Joint test F	4.29*** (5, 296 df)	5.09*** (5, 180 df)	3.94** (5, 281 df)	5.14*** (5, 173 df)
Immigrant's birth area
Africa	1.153	3.206	1.428	3.033
China	1.619	1.430	1.604	1.699
India	1.838	2.268	1.819	2.419
Philippines	0.797	4.624	0.745	3.513
Vietnam	1.369	1.968	1.475	2.211
Other Asia	1.100	1.713	1.092	1.882
Oceania	2.359	1.984	2.384	2.097
Dominican Republic	3.797	2.108	3.685	2.012
El Salvador	0.662	2.237	0.350	2.177
Guatemala	1.557	1.540	1.405	1.524
Haiti	2.215	2.572	2.050	2.441
Mexico	1.002	2.533	0.840	2.499
Other North/Central Am.	0.733	2.256	0.705	2.151
Colombia	1.204	3.059	1.089	2.836
Other South America	1.679	1.532	1.393	1.686
Joint test F	3.18*** (15, 296 df)	4.54*** (15, 180 df)	2.91*** (15, 281 df)	4.35*** (15, 173 df)
Constant	−0.987 (0.82)	−6.479 (3.54)***	−0.808 (0.68)	−5.818 (3.18)**
R-squared	0.248	0.318	0.252	0.310
Observations	320	204	305	197

Notes: Skin color is measured by the interviewer on an 11-point scale. NA = not applicable. Robust standard errors; absolute values of asymptotic t-ratio appear in parentheses under parameter estimates for numeric and binary variables. Joint tests reported for multiple-category categorical variables.

^*p < .05,

^**p < .01,

^***p < .001;

two-tailed tests for single coefficients, one-tailed for joint tests

Sponsor's age and age-squared are jointly statistically significant only in the women's equations, and they indicate that the propensity to marry darker increases until about age 38 and subsequently declines. This result is consistent with an interpretation in which reaching out across the color line peaked for native-born U.S. citizen women born in 1965 – at the height of the civil rights movement 2013; suggesting integrationist parental influences during the formative years.

The race-ethnicity regressors are jointly statistically significant, with nonHispanic whites marrying darkest. Immigrant birth area is also statistically significant. Skin-color difference is greatest when the immigrant hails from the Philippines, Africa, and Colombia (among female sponsors) and from the Dominican Republic, Oceania, and Haiti (among male sponsors).

7.5. Spouse-of-U.S.-Citizen Couples Contrasted with Other Married Couples

Within-couple diversity is clearly not trivial among couples involving a native-born U.S. citizen sponsor of a spouse, especially if the sponsor is female. But how different are these couples from the other couples in the 2003 immigrant cohort? To address this question, Table 23 reports the skin-color difference measured as husband's skin color minus wife's skin color, for all the visa classes except those designated for unmarried persons (children of U.S. citizens, family first preference for unmarried sons and daughters of U.S. citizens, and the subset of family second preference for unmarried sons and daughters of LPRs).

Table 23.

Skin-Color Difference Among Married Couples, by Sampled Immigrant's Visa Class and Sex: NIS-2003 Cohort

Visa Class	Men		Women
	Skin-Color Difference	N	Skin-Color Difference	N
Spouse of NB U.S. citizen	1.566	106	−.651	133
Spouse of FB U.S. citizen sponsor	.269	104	.267	193
Parent of U.S. citizen	.265	116	.194	97
Adult married child of U.S. citizen	.315	31	.363	25
Spouse of adult child of U.S. citizen	.0527	42	.273	37
Sibling of U.S. citizen	.302	43	.271	45
Spouse of sibling of U.S. citizen	.132	29	.541	31
Spouse of LPR	.590	18	.761	56
Employment principal	.209	217	.257	64
Spouse of employment principal	.150	31	−.0766	58
Diversity principal	.887	117	.0560	124
Spouse of diversity principal	.390	43	.674	32
Refugee/asylee/parolee principal	.217	94	.453	42
Spouse of RAP principal	.450	16	−.0116	31
Legalization principal	.211	121	−.0592	106
All married respondents & spouses	.487	1139	.0659	1078

Notes: Skin color is measured by the interviewer on an 11-point scale. Skin-color difference is defined as husband's skin color minus wife's skin color. Estimates based on weighted data.

If respondents marry within their ancestral group, all the skin color differences in Table 23 would be positive – that is, husbands would be darker than their wives. The averages in Table 23 would cluster around some small positive number. Of course, it is possible that this “natural” gender difference in skin color may differ across ancestral group, such that the skin color difference between husbands and wives would be larger in some ancestral groups than in others. To the extent that visa categories differ in their origin-country composition, the natural skin color difference would vary across visa categories. Moreover, our ability to discern this natural skin color difference depends on interviewer behavior. That is, it depends on interviewers coding the skin color they see and not mentally adjusting for the sex difference and assigning husbands and wives with similar origins the same skin color score.

Skin color differences larger than the natural difference indicate that husbands are darker than expected (or, equivalently, wives lighter), and skin color differences smaller than the natural difference indicate that husbands are lighter than expected (or, equivalently, wives darker). Even if the natural difference is unknown, negative magnitudes indicate that husbands are lighter than expected (or, equivalently, wives darker). As shown, the two extreme skin color differences in Table 23 remain those involving native-born U.S. citizen sponsors of spouses, namely, 1.566 among women sponsors and −.651 among men sponsors. There are three other negative differences, among marriages involving women legalization principals and wives of employment and refugee/asylee/parolee principals. These skin color data can thus be used to explore variation in the gender difference in skin color across ancestral groupings, using NIS information on country of birth, religion, and language – and, of course, to distinguish the natural difference from skin color difference induced by intermarriage across ancestral groupings.

The information in Table 23 can also be used as a springboard to explore the characteristics of the missing spouses – that is, spouses who refused to be interviewed – of main sampled immigrants in the subset of visa categories in which visas are available for accompanying spouses of principals. To see this, recall that the main sampled immigrants include both immigrants who received visas as principals and immigrants who received visas as accompanying spouses (Table 3). Accordingly, for each of the visa categories which permit accompanying spouses, NIS data have information based on two subsets: (1) sampled-immigrant principal and spouse, and (2) sampled-immigrant spouse and spouse-principal. Average skin color difference should be the same in both subsets. Under the assumption that the decision by main sampled immigrants to participate is unrelated to skin color, any discrepancy in average skin color difference across the two subsets indicates missing spouses with the missingness related to skin color. Note that the converse does not hold. If the two subsets have equal average skin color difference, they could still have missing spouses, but in this case, the missing spouses would be complementary with respect to skin color.

To illustrate, consider immigrant couples in which the main sampled immigrant is a married son of a U.S. citizen (that is, a male obtaining LPR as a principal in the married-child-of-U.S.-citizen visa category); his wife agreed to be interviewed, and both spouses received a skin color score. As shown in Table 23, in such couples the average skin color difference is .315. Under the assumption that the decision to participate in the survey is unrelated to skin color among both sons and daughters-in-law drawn as main sampled immigrants, these couples' skin color difference should be the same, except for sampling fluctuation, as that among couples in which the main sampled immigrant is the daughter-in-law of a U.S. citizen (that is, the wife of a principal in the married-child-of-U.S.-citizen visa category). And, indeed, average skin color difference is almost the same – viz., .273 for the second subset.

These subsets of couples can be visually discerned as they occupy “diagonals” on Table 23. To continue the illustration, consider now immigrant couples in which the main sampled immigrant is the married daughter of a U.S. citizen (that is, a female obtaining LPR as a principal in the married-child-of-U.S.-citizen visa category). The skin color difference in such couples is .363. However, the skin color difference in the second subset – couples in which the main sampled immigrant is the son-in-law of a U.S. citizen (that is, husband of a principal in the visa category) – is .0527. The discrepancy between .363 and .0527 suggests that one or both of the two subsets has missing spouses. If the true natural skin color difference in this visa category is .363, then the observed skin color difference of .0527 in the subset with the husband main sampled immigrant is too low, indicating that the missing wives are lighter skinned. But if the true natural skin color difference is .0527, then the observed skin color difference of .363 in the subset with the wife main sampled immigrant is too high, indicating that the missing husbands are lighter skinned.

It is outside the scope of this paper to analyze the skin color of the missing spouses, but it is evident both that NIS data can shed light and that this information can be used to shed light on other behavioral patterns observed in the data.

As a final way of assessing contributors to skin color diversity among married couples, we obtain the absolute skin-color differences, ignoring any directionality by sponsor or by marital partner. These figures (not shown) indicate that the three largest average absolute differences are among native-born citizens who sponsor spouses (2.014 and 1.28, among female and male sponsors, respectively) and legalization principals (1.279). The potentially exemplary character of native-born U.S. citizen women, who are marrying darker, remains unchallenged. It is noteworthy that the current American President's native-born white mother was in the vanguard, marrying a black from Africa.

Of course, much further research is warranted, including a deeper look at spouse selection among both native-born and foreign-born U.S. citizens, relations between the spouses, effects on gender inequality, and, as future rounds of the NIS accumulate, the unfolding of marital cohesiveness and disruption.

8. ENGLISH SKILL AMONG YOUNG CHILDREN AND THEIR IMMIGRANT PARENTS

8.1. Learning a New Language

Perhaps no single characteristic is as emblematic of both the life chances of immigrants and their children and the reactions of natives as English fluency (Alba and Nee 2003; Bean and Stevens 2003; Jasso and Rosenzweig 1990; Portes and Rumbaut 2006; Rumbaut, Massey, and Bean 2006). Yet for adults, learning English can be difficult, especially if first encountered after adolescence, if there is little time left after work, and if there are few English-class options. Moreover, immigrants who have spent time as illegals may experience even more difficulty learning English, given that perhaps the safest way to avoid detection may be to blend in with co-nationals.

Meanwhile, however, the children of immigrants quickly find English in their environment. Even when parents know no English or, even if they know English, decide to use only their first language at home – so that the children can master the origin language, while hopefully also mastering English in school – their children, once they are in school, face an environment rich in English resources. Mass media, labels on food at the grocery store, toy boxes at the toy store, signs and billboards – all envelop young children in English. Not surprisingly, the evidence points in the direction of fairly rapid acquisition of English among the second generation (Alba and Nee 2003; Bean and Stevens 2003; Portes and Rumbaut 2006; Rumbaut, Massey, and Bean 2006).⁴¹

In general, there are four distinct sets of children whose life chances are affected by international migration. The four sets, formed by crossclassifying children's country of birth and country of residence, are: (1) foreign-born children residing with their foreign-born parents in the destination country; (2) native-born children of foreign-born parents in the destination country; (3) foreign-born children whose parents are in the destination country but who are left behind in the origin country; and (4) native-born children whose parents are in the destination country and send the children to be raised in the origin country. The NIS enables a close look at all four sets.

In this section, we use NIS data to examine the influences which militate for and against achieving fluency in English among parents and two sets of children – foreign-born and native-born residing with their parents in the United States. Influences to explore include the origin-country and childhood environment as well as situational factors captured by visa category and previous illegal experience.⁴²

Among children, two potentially important factors are whether the child was born in the United States and, if not, the age at entry. With few exceptions – notably children of foreign diplomats -- children born in the United States acquire U.S. citizenship at birth and enjoy the full set of rights of native-born U.S. citizens – including the constitutional right to run for the Presidency. Foreign-born children of immigrants, including those brought in as infants or toddlers -- sometimes called the 1.5 generation -- have no claim on a life in the United States unless they acquire legal status as the minor children accompanying their parents or are sponsored for LPR by their parents. If their parents do not obtain legal status, the children are deportable. One may speculate that both the children themselves and their parents are more likely to invest, and to invest more heavily, in the U.S.-specific capital – including English language skill – of native-born children, who have every claim on a full American life. Of course, given the pervasiveness of English in the environment, such a nativity premium may be small (though possibly larger for other skills and behaviors, such as reading about American history and visiting the country's foundational monuments).

Another factor potentially influencing children's English fluency is the presence of native-born persons in the immediate environment. If the child's immigrant parent is the spouse of a native-born U.S. citizen (see Table 4), the household includes a native-born adult. This adult (married to the child's immigrant parent) may in fact be the child's own biological parent; or this adult may be the child's step-parent. Under both scenarios, the household may include native-born children who are the immigrant child's siblings. In the scenario with the step-parent, the household may also include native-born half-siblings and step-siblings. A natural question to ask is whether these scenarios have different effects on the child's English fluency. A priori one may speculate that if the child is the biological child of the native-born U.S. citizen sponsor, the child may have more English in the environment. However, a countervailing speculation is that parents may seek to diversify their children's language portfolio and discourage English at home, analogously to diversifying the family citizenship portfolio, as has been empirically found (Jasso and Rosenzweig 1990).

8.2. Special NIS Data for Studying English Fluency Among Immigrant Parents and Their Young Children

The New Immigrant Survey, besides interviewing the main sampled immigrants in the Adult Sample and their spouses (Table 1), also interviewed up to two randomly selected children age 8–12 residing in the household. A total of 1,072 children were thus interviewed in the Adult Sample. These children include biological, step, and adopted children. This section focuses on interviewed biological children of the main sampled immigrant, of whom there are 1,014. These are children of 887 main sampled immigrants.

Children were given the same broad choice of interview language as their parents, and the data provide information on whether English, if chosen, proved to be the only language used by the child during the interview. Accordingly, we define for the children the same behavioral measure of English fluency used in Section 6 for adults. The children are coded “1” if the entire interview was completed solely in English.

The NIS also obtained information on the children's first language, asking, “What was the first language you learned to speak when you were a child?” We coded the responses into three categories. These and their proportions among the biological children are: English, 16.5%; Spanish, 48.2%; and other, 35.1%.⁴³

To characterize the child's nativity and age at entry we combine nativity and age at entry to classify children into three categories: (1) born in the United States; (2) born abroad, entered before age four; and (3) born abroad, entered at age four or older. The first two categories represent an initial crude attempt to distinguish between native-born children of immigrants and those brought in as infants or toddlers, as discussed above. The proportions in these categories among the biological children are: born in the United States, 45.1 percent; entered before age four, 6.41 percent; entered at four or older, 43.1 percent; missing, 5.4 percent.

To represent more sharply the environment associated with the spouse-of-NB-citizen visa category, we replace it by two binary variables, indicating whether the immigrant parent's spouse is or is not the biological parent of the focal child. The proportion with a native-born parent is 71.2 percent.⁴⁴

8.3. Parental Origin Country, Previous Illegal Experience, and English Fluency among Parents and Children

We begin by looking at the class-of-admission and origin-country distributions of the children's parents. By far the largest visa classes of the entire cohort are the two spouse-of-U.S. citizen categories (16.2% and 17.9% for NB and FB sponsors, respectively), followed by parent of U.S. citizen (11.9%), and legalization (7.98%). In contrast, among the parents of young children age 8–12, the largest visa category is that of legalization (at 22.1% almost three times as large as among the cohort as a whole), followed by spouse of FB citizen (11.7%), spouse of NB citizen (7.34%), and spouse of LPR (7.02%). As would be expected, the parent-of-U.S.-citizen category is not among the top categories. Among the children, the figures are similar, except that spouse of NB citizen falls to fifth place (6.95%), spouse of LPR ascends to third place (7.75%), and refugee/asylee/parolee principal rises to fourth place (7.21%). These early numbers hint at the intensity of parental previous illegal experience in these children's histories.

Turning to parental origin country, recall from Section 2.3 that the top five birth countries for the adult immigrants are Mexico (17.5%), India (7.30%), El Salvador (6.11%), Philippines (5.47%), and China (5.27%). The origin-country distribution of the children's parents is likely to differ from that among all adult immigrants, given that country-specific immigrant streams may differ in age. For example, while only 2.69% of those born in El Salvador immigrated as parents of U.S. citizens, 23.3% of those born in China did so.

Indeed, the parental origin-country distribution among the biological children replaces China with Guatemala in the top five, alters the ordering, and yields substantially larger proportions from Mexico and El Salvador: Mexico (25%), El Salvador (13.9%), Guatemala (6.58%), India (5.19%), and Philippines (4.79%). In sixth through tenth place are: Vietnam (3.71%), Dominican Republic (3.04%), Cuba (2.51%), China (1.95%), and Ukraine (1.83%).⁴⁵

The top three parental origin countries are the same three countries with the highest previous illegal experience among the adult immigrants, with El Salvador in first place, followed by Guatemala and Mexico (Section 2.4). Among the children, the estimated proportions with parental previous illegal experience range from the lower-bound estimate of 28.6% (counting immigrants with either a legalization immigrant visa or an EWI/WI nonimmigrant code) to 48.9% (including all the information on the immigrant record) to 52.3% (adding as well the survey measures). Among children whose immigrant parent is from one of the top three countries, the estimated ranges of parental previous illegal experience are: Mexico (36–87.4–91.5%), El Salvador (86.5–94.9–95.6%), and Guatemala (88.5–92–97.4%). Again, the figures change only slightly if calculated on the set of parents; for example, the three overall proportions, starting with the lower bound, are 28.6, 49.2, and 52.6%. Thus, immigrants with co-resident children aged 8–12 are more likely to have previous illegal experience than the adult immigrants as a whole, for whom the three corresponding figures are 11.4, 35.7, and 39.6% (Table 6). And over half the children aged 8–12 have parents with illegal experience.

The proportion fluent in English among the 887 immigrant parents is estimated at 26.0% (substantially less than among all the main sampled immigrants, for whom it is 41.2%), and it varies greatly across birth country, visa class, and previous illegal experience. Among the top ten parental origin countries, it ranges from lows of 2.25% and 4.66% (Vietnam and Ukraine, respectively) to highs of 55.5% and 69.0% (India and Philippines, respectively); the proportions for Mexico, El Salvador, and Guatemala are 6.55%, 10.1%, and 11.3%, respectively. By visa class, parental English fluency ranges from lows of 9.31% and 13.4% among legalization principals and spouses of LPRs, respectively, to highs of 59.6% and 63.9% among employment principals and spouses, respectively. And English fluency differs across immigrant parents with and without illegal experience. Among new arrivals and adjustees with a valid visa code, who additionally have no history of illegal experience as measured by the survey questions, the estimated proportions fluent are 34.1% and 55.6%, respectively, while among adjustees with a legalization visa or an EWI nonimmigrant code, the proportion is 8.55%, and among all adjustees with previous illegal experience, the proportion fluent is 13.4%.

The situation is quite different among the children. The proportion who completed the interview exclusively in English is 68% – more than twice as high as among their immigrant parents (26.0%). With respect to parental origin country, the proportion fluent in English ranges from lows of 21.7% (Vietnam) and 23.0% (China) to highs of 87.2% (India) and 96.2% (Philippines). For the top three parental birth countries, the estimates are 67.2% (Mexico), 70.1% (El Salvador), and 71.7% (Guatemala) – dramatically larger than the 6.55%–11.3% among their parents. Across parental visa class, the proportions of children fluent in English range from 37.6% and 41.5% (sibling principals and spouses, respectively) to 84.6% and 80.4% (employment principals and spouses). Across parental illegal experience, the children of adjustees are almost indistinguishable from each other regardless of their parent's illegal experience – with proportions fluent that range from 71.5% among children with an EWI parent to 81.2% among children of a valid-visa parent with no illegal experience on the survey measures – while the children of new-arrival parents with no illegal experience register a proportion fluent of 55.3%.

Among immigrants with no estimated previous illegal experience, both adjustees and their children have greater English fluency than new arrivals and their children, by 20–25 percentage points. However, among adjustees with illegal experience, the parents have very low English fluency while their children's English fluency rivals that of the children of adjustees with no illegal experience.

Taken together, these patterns suggest two potentially important dynamics, first, a leveling across generation – with children far more similar to each other than are their parents – and, second, a kind of compensation for parental illegal experience – with children of previous illegals achieving great English fluency, possibly in response to the hardships their parents endure or to practice translating and interpreting for their parents (Valdés 2003).

8.4. Multivariate Analysis of English Fluency Attainment among Parents and Children

Table 24 reports binary logit estimates of the determinants of English fluency among the biological children aged 8–12 interviewed in the NIS and their immigrant parent. One equation is estimated for the parent and two for the children, one of which parallels closely the parent equation while the other adds parent's English fluency as a regressor. All three equations include the binary variable indicating whether English is an official or dominant language of the parent's country of birth. The parent's equation includes the set of visa categories which, as in previous sections, distinguish between principals and spouses (see Tables 1, 3, 4, etc.), who may differ in knowledge of English. The children's equations, however, group together principals and spouses from the same visa category, given that the child's environment includes both parents. As well, the children's equations replace the category for spouse of native-born citizen with the two categories described above distinguishing between the subsets in which the immigrant parent's spouse is or is not the child's biological parent. The origin area dummies are for the top ten parental countries (Section 8.3). The children's equations also include the new variable characterizing the child's nativity/entry. Finally, the children's equations include a correction for heteroskedasticity due to the clustering of children within family.

Table 24.

English Fluency Among Children Age 8–12 and Their Immigrant Parents: Logit Estimates, NIS-2003 Cohort

Regressor	Parents	Children
		(1)	(2)
Sex (1 = female)	−0.699 (3.24)***	0.121 (0.74)	0.0631 (0.38)
Age	0.0174 (0.93)	0.159 (2.69)**	0.142 (2.36)*
Spouse of NB citizen	1.158	NA	NA
Biological child of U.S. citizen	NA	−.0860	.0383
Not biological child of U.S. citizen	NA	.937	.956
Spouse of FB citizen	.406	−.211	−.295
Parent of U.S. citizen	−.0474	1.283	1.269
Adult unmarried child of U.S. citizen	.608	−.529	−.495
Married child of U.S. citizen	1.345	−.302	−.443
Spouse of adult child of U.S. citizen	.0544
Sibling of U.S. citizen	−.633	−.702	−1.569
Spouse of sibling of U.S. citizen	−.555
Spouse of LPR	.162	−.261	−.238
Child of LPR	.0108	−.0437	.0709
Employment principal	1.463	.243	−.0954
Spouse of employment principal	1.451
Diversity principal	.280	−.0511	−.0553
Spouse of diversity principal	−.345
Refugee/asylee/parolee	.620	−.761	−.987
Spouse of refugee/asylee/parolee principal	.659
Joint test chi2	32.68** (16 df)	21.56* (12 df)	20.72 (12 df)
Adjustee	1.001	1.625	1.518
Previous illegal experience	−.633	.203	.429
Joint test chi2 (2 df)	6.78*	25.77***	23.69***
Childhood language Spanish	−3.261	−1.664	−1.500
Childhood language English & Spanish	−2.442	NA	NA
Childhood language English & other	−.969	NA	NA
Childhood language Spanish & other	−2.269	NA	NA
Childhood language other	−2.168	0.224	.289
Joint test chi2	23.01*** (5 df)	24.16*** (2 df)	25.75*** (2 df)
Ukraine	−2.032	−1.338	−1.000
China	−1.407	−1.766	−1.674
India	−0.903	.742	1.087
Korea	−3.290	−1.485	−.865
Philippines	−0.840	.742	1.057
Vietnam	−2.287	−1.361	−1.071
Cuba	−1.361	−1.084	−.743
Dominican Republic	−1.419	−0.878	−0.672
El Salvador	−0.462	−0.983	−0.639
Guatemala	−0.647	−1.094	−0.775
Mexico	−.997	−0.971	−0.573
Joint test chi2 (11 df)	32.61***	24.52*	16.90
English official language of parental origin country	2.115 (3.23)***	1.661 (3.21)***	.923 (1.71)
Child entered U.S. at age 4+	NA	−.268	−.0203
Child entered U.S. under age 4	NA	.804	.970
Child born in U.S.	NA	1.005	1.108
Joint test chi2 (3 df)	NA	13.21**	10.89*
Parent interviewed in English	NA	NA	1.735 (5.05)***
Constant	−.547 (.47)	−.887 (.97)	−1.504 1.69)
Log pseudolikelihood	−308.61	−460.98	−440.19
Observations	870	998	998

Notes: NA = not applicable. Standard errors in child regressions corrected for heteroskedasticity due to clustering within family. Absolute values of asymptotic t-ratio appear in parentheses under parameter estimates for numeric and binary variables. Joint tests reported for multiple-category categorical variables.

^*p < .05,

^**p < .01,

^***p < .001;

two-tailed tests for single coefficients, one-tailed for joint tests

The first result is that children do not reproduce the adult gender effect. While immigrant mothers (and immigrant women in general) are statistically significantly less likely to be fluent in English than immigrant fathers (and immigrant men in general), there is no statistically discernible gender effect among the children. Indeed, the point estimates indicate a greater English fluency among girls.

The estimates for parental visa category and the adjustee and illegal experience variables echo some of the results in the previous section and introduce others. Among the parents, the top four visa categories associated with English fluency are the two employment categories, the category for adult married children of U.S. citizens, and the category for spouses of NB citizens. Among the children, those in the top visa category have an immigrant parent who attained LPR as the parent of a U.S. citizen – that is, children who have siblings or half-siblings who are U.S. citizens. The second top visa category includes children whose immigrant parent is married to a native-born U.S. citizen who is not the biological parent of the focal child – children, that is, who may have half-siblings or step-siblings who are U.S. citizens. These results, together with the rather low probability of English fluency among children whose biological parent is a native-born citizen, suggest three possible dynamics: (1) parents may seek to diversity their children's linguistic portfolio, as noted above; (2) children learn from their environment -- learning from a native-born adult in the household who is not their biological parent; and (3) siblings, half-siblings, and step-siblings may play an important role in promoting English fluency among children.⁴⁶

The adjustee and previous illegal experience variables are jointly statistically significant, highly so in the children's equations. Other things the same, adjustees are more likely to be fluent in English, as one would expect given their longer time in the United States. However, parental previous illegal experience operates in opposite ways for parents and children, echoing the raw proportions discussed in the previous section. While parents are less likely to be fluent in English if they have previous illegal experience, children do best precisely when their parents have illegal experience. This effect strongly suggests that children of illegal immigrants are transcending their origins, whether the mechanism involves compensating for the hardships endured by their parents and/or becoming “gifted interpreters” for their parents, to use Valdés' (2003) evocative phrase.

Note that there is a highly statistically significant positive effect of parental English fluency on child fluency (in the children's second equation). Thus, even when one of the parents is fluent in English, the child still does better when that parent is legalizing from an illegal status or has been illegal in the past, suggesting a keenness to achieve and/or the presence in the household of other relatives (including the other parent) for whom to translate and interpret.

The children are not only learning from their environment, but also they are learning more and more with each passing year, as indicated by the statistically significant age effect. There is no age effect, however, for their parents (with or without a square term).

Of course, origin country and childhood language matter, for both parents and children. The binary variable for English an official/dominant language has a highly statistically significant positive effect on parents – exactly as in the equation for black immigrants. In the children's equations, however, the effect, though positive, reaches statistical significance only in the first equation. Moreover, the magnitude declines in the second equation – which includes the binary variable for parental English fluency – indicating that the origin-country linguistic environment is important for the parent but not directly for the children.

As for parental origin country, the birth areas most conducive to English fluency, net of the English official/dominant language variable, are India and the Philippines, for children, and El Salvador and Guatemala for the parents. Four of the bottom five countries for both parents and children, albeit in different order, are Ukraine, China, Korea, and Vietnam. Also in the bottom five are Guatemala children and Dominican Republic for parents.

Net of parental origin country and whether English is an official or dominant language there, children are least likely to be fluent in English if their first language was Spanish; parents are least likely to be fluent in English if their languages at age 10 included Spanish, whether alone or in combination with English or another language. Thus, Spanish continues to exert downward pressure on English fluency (consistent with it lasting longer than other immigrant languages, as found by Rumbaut, Massey, and Bean 2006).

The nativity/entry variables indicate that, as expected, children born in the United States or brought to the United States under the age of four are more likely to be fluent in English than children brought in at older ages (or children with missing information on nativity and entry). The results also support an estimate of the nativity premium – small but positive. Further research is warranted to explore more deeply the nativity premium.

These results add to the accumulating evidence that immigrant children and children of immigrants quickly adopt English (Alba and Nee 2003; Bean and Stevens 2003; Portes and Rumbaut 2006; Rumbaut, Massey, and Bean 2003). Moreover, they signal the possibility that children may overachieve to overcome parental disadvantages (Bean and Stevens 2003), with children of previously illegal parents potentially outperforming children of parents who never lapsed into illegality.

Information obtained during the NIS second round should prove illuminating. Whether parents (and the other adults) improve their English, whether children continue on this achieving course, whether determinants and correlates shift their patterns – these will be possible to assess. It will also be important to track the children's early occupational experiences and their effects on educational and socioeconomic outcomes (Mortimer 2003). As well, further research is warranted on the subset of children whose immigrant parents are spouses of U.S. citizens (complementing the work here), exploring not only family language dynamics and whether there are specific patterns involving child's gender and parent's gender but also exploring step-siblings – sponsor's children from previous unions -- and the part they play in socializing and integrating the newcomer children. It is possible that step-siblings may join adoptive parents, native-born U.S. citizen women, and black immigrants from Africa as special protagonists in eradicating inequalities of various kinds.

9. CONCLUDING NOTE

Stratification is embedded in migration, and, increasingly and around the world, migration is embedded in stratification. One day soon it will be impossible to understand one without the other. This paper has used a unique new data set, the New Immigrant Survey, to explore the connections between migration and stratification in six dimensions that exemplify the parts played by government, private citizens, and immigrants and their children in the unfolding of the immigrants' and their children's life chances and the impacts on the stratification structure of the United States.

The paper made several methodological and substantive contributions. Methodologically, the paper provides basic information about the U.S. immigration context, necessary for understanding the immigrants' lives and the rich new detail in the NIS; uses a new skin color scale to understand not only skin color among new immigrants but also marital diversity in skin color and nonresponse to the standard race question used in U.S. surveys; uses whole-distribution graphical tools whose usefulness is increasingly appreciated; conducts Kolmogorov-Smirnov tests on the skin color distributions; and explores difference-in-differences procedures for inferring the skin color of missing (noninterviewed) spouses.

Substantively, the paper provides a range of new results with potentially important implications for both science and policy. First, starting with an examination of lost documents in immigration offices, we found that documents are more likely to be lost in offices of the INS/USCIS in the United States than in consular or embassy posts overseas, and that the characteristics of immigrants and their sponsors affect the probability of lost documents – for example, in the immigration of spouses of U.S. citizens, both stateside and abroad, documents are more likely to be lost if the U.S. citizen sponsor is foreign-born than if the sponsor is native-born. Second, lost documents increase the probability that the immigrant experienced depression as a result of the visa application process; adjustee immigrants are more likely to experience visa depression; and only certain kinds of kin protect against visa depression. Third, men are more likely than women to declare that they are the principal applicant in the case, whether or not they are, raising questions about gender entitlement and the possibility of gender effects in reported schooling and earnings in survey data. Fourth, immigration is increasing diversity in the U.S. population – by introducing substantial contingents of Asian and Hispanic immigrants – and threatens to overturn racial and skin color associations with skill – by introducing accomplished black immigrants from Africa. Fifth, native-born U.S. citizen sponsors of spouses are marrying spouses who are darker than themselves, and this is most pronounced among female sponsors, suggesting that they are in the vanguard of increasing marital diversity in skin color. Sixth, young children are dramatically more likely than their immigrant parents to be fluent in English; there is a nativity premium, such that children born in the United States are more likely to be fluent in English than children brought before age four; and children of previously illegal immigrants are especially more likely to be fluent in English, suggesting that they may be compensating for the hardships they have seen their parents endure and/or that they have obtained practice translating for their parents or other family members.

Much more will be learned about these processes and their longer-term effects, as further analyses are carried out on the first and second round data and new information is collected in subsequent rounds of the NIS. It will be possible to assess whether lost documents and visa depression engender diminished attachment to the United States, visible in emigration, naturalization, and voting, and whether visa depression affects subsequent health outcomes. As well, it will be possible to observe whether and how new legal immigrants shed the habits of their former life, in some cases shedding the habits of illegality, in others the habits of elitism and gender entitlement. And it will be possible to gauge whether the auspicious signs of highly accomplished black immigrants and of marital diversity in skin color develop into social forces. As well, it will be possible to see whether other aspects of immigration, beyond those analyzed here, contribute to intensifying or attenuating the stratification structure of the United States.

Immigration research and NIS data are as Shakespeare imagined Cleopatra: they make hungry where most they satisfy.

Table A.1.

Basic Survey Characteristics of Adult Sample: NIS-2003

Characteristic	Men	Women	All Immigrants
Total interviewed in Adult Sample	4124	4449	8,573
Not overseas	3992	4260	8,252
Interviewed in person	1632	1797	3,429
Interviewed by phone	2484	2636	5,120
Average number of months between LPR & interview	3.91	3.91	3.91

Notes: Figures represent raw cases with no sampling weights. Overseas cases were administered an abbreviated questionnaire, and their spouses were not interviewed.