Legal Status and Health Care: Mexican-Origin Children in California, 2001–2014

Abstract

Using restricted data from the 2001–2014 California Health Interview Surveys, this research illuminates the role of legal status in health care among Mexican-origin children. The first objective is to provide a population-level overview of trends in health care access and utilization, along with the legal statuses of parents and children. The second objective is to examine the nature of associations between children’s health care and legal status over time. We identify specific status-based distinctions that matter and investigate how their importance is changing. Despite the continuing significance of child nativity for health care, the descriptive analysis shows that the proportion of Mexican-origin children who are foreign born is declining. This trend suggests a potentially greater role of parental legal status in children’s health care. Logistic regression analyses demonstrate that the importance of parental legal status varies with the health care indicator examined and the inclusion of child nativity in models. Moreover, variation in some aspects of children’s health care coalesced more around parents’ citizenship than documentation status in the past. With one exception, the salience of such distinctions has dissipated over time.

The cumulative impact of immigration to the United States over the past several decades has been substantial. Approximately 35 million immigrants have acquired permanent resident status since 1970, including nearly seven million Mexicans (Department of Homeland Security 2013). Mexico is also the origin of seven million unauthorized immigrants (Hoefer, Rytina and Baker 2011; Passel and Cohn 2011). Thus, the number of undocumented residents is equivalent to the number obtaining green cards.

According to the American Community Survey, the Mexican-origin population comprises 11% of the national total. However, it is distributed unevenly across states. California accounts for over one-third of both foreign- and native-born Mexican-origin persons, and one-fourth of the undocumented (Hoefer, Rytina and Baker 2011). In addition, nearly half of all pre-teen children in the state are of Mexican descent. Such figures suggest that migration and family circumstances intersect for a large number of children in California.

This intersection is potentially heightened by the federal government’s reliance on deportation, surveillance and coordination with local law enforcement to deal with unauthorized immigrants. These policy tools are portrayed by some as part of an anti-immigrant shift that fuels an inhospitable social climate through “scapegoating, ostracism, family and community separation” (Kanstroom 2007: 5; also Chavez 2014). Indeed, an alleged consequence of changes in public policies and sentiments is that legal status distinctions have “hardened” with growing differences in the treatment of naturalized citizens, permanent residents and the undocumented (Coutin 2011).

The intensification of legal status distinctions is pertinent to the children of immigrants because parents have status-related resources that shape the life chances of their offspring, potentially through pathways that involve health care. For example, many undocumented residents are impoverished and ineligible for public programs that serve the low-income population. Such disadvantages have potential implications for their dependents, including those who are eligible for public programs as U.S.-born citizens. Undocumented parents may limit children’s contact with health care providers because of fear of apprehension, lack of knowledge, lack of financial resources and “daunting encounters with a fragmented, bewildering, and hostile system” (Okie 2007: 525). In short, barriers may have become more formidable with the intensification of legal status distinctions for parents, regardless of the nativity of children themselves.

Using the 2001–2014 California Health Interview Surveys (CHIS), this research illuminates the implications of legal status for health care among Mexican-origin children. The first objective is to provide population-level evidence of trends in health care access and utilization, as well as the legal statuses of children and their parents. The second objective is to identify how legal status affects children’s health care, with particular attention to temporal changes in the role of parental status. The availability of rare information on detailed legal status distinctions in these surveys provides leverage to determine whether differences have intensified or dissipated. This objective emphasizes the legal status of parents rather than children (albeit not exclusively) due to major changes in population composition that are revealed in the descriptive analysis. Nonetheless, insights on the consequences of having two undocumented parents for both native-born and undocumented children are provided. Those in the former category have captured a significant amount of interest for their potential vulnerability as members of “mixed status” families (e.g. Fix & Zimmermann 2001).

BACKGROUND

The terms legal status and immigration status are used interchangeably to denote hierarchically-organized categories established by institutions through laws, rules and procedures that regulate access to the rights of membership in society. Naturalized citizens are positioned at the apex of this hierarchy among the foreign born. Citizenship is available to authorized immigrants who demonstrate their commitment to the United States. This requires evidence of permanent settlement, knowledge of English and civics, compliance with laws and “good moral character” along with an oath of allegiance (U.S. Citizenship and Immigration Services 2012: 1). Lawful permanent residents (LPR’s) occupy the middle position. Issued renewable “green cards” that must be carried at all times, they are authorized to live and work in the United States. LPR’s are also entitled to public benefits after a probationary period. At the bottom are undocumented migrants who lack permission for permanent or temporary residence.

The salience of these categorical distinctions must be problematized to assess how institutions structure life chances, particularly given the tendency to bifurcate statuses based on “essential differences” for comparisons. Indeed, there is little consensus on which groups are essentially similar or different (Oropesa, Landale and Hillemeier 2015). Bifurcation according to documentation status (documented vs. undocumented) equates permanent residents with naturalized citizens based on their unrestricted rights to employment and residence. Bifurcation according to citizenship places permanent residents with the undocumented to identify non-citizens, a distinction that is allegedly fundamental for life chances (Bosniak 2006; Coutin 2011). This classification is consistent with claims that gaps among segments of the non-citizen population have narrowed with the ratcheting up of deportation, anti-immigrant sentiment and stigmatization. Nonetheless, the relative salience of various legal status distinctions is a matter for empirical research.

Understanding how legal status affects health care requires attention to access. Andersen and Newman (2005: 8) define access as “the means through which the patient gains entry to the medical care system;” that is, direct payment for services versus payment by “government, voluntary health insurance or other third party payers.” The inability to cover expenses by these means is a major access-related barrier to care (Aday and Andersen 1974).¹ This barrier is especially imposing for the undocumented, who may forego care because financial constraints undermine their ability to afford private insurance or direct payment for services. Access to government support may also be blocked because public insurance programs are generally limited to low-income citizens and permanent residents with at least five years in the United States (Broder 2005).

In California, some barriers have been lowered despite the restricted eligibility of the undocumented for Medicaid (Medi-Cal) and the Children’s Health Insurance Program (SCHIP/Healthy Families) in the past.² Starting in 2001, county-based initiatives mobilized public and private resources to cover undocumented children under the banner of the Healthy Kids Program (Stevens, Rice and Cousineau 2007). Coverage also expanded in some counties under the Kaiser Permanente Child Health Plan and CalKids, a program sponsored by various organizations. Healthy Kids coverage expanded from 7 counties in 2004 to over 20 counties by 2007, before contracting to 15 by 2012. Currently, 22 counties are devoid of programs, 10 have one program, 18 have two programs and 8 counties participate in all three (Cousineau et al. 2012).

A case could be made that attention to institutional barriers to public benefits is misplaced because most Mexican-origin children are native born. Yet, the National Health Interview Survey (NHIS) indicates that the Mexican-origin child population fares relatively poorly on some dimensions of health care access and utilization (Perez et al. 2009). Though similar to non-Hispanic Whites in having a usual source of care and use of emergency rooms, they are less likely to visit a doctor and more likely to delay care. Of course, such findings are difficult to assess without information on legal status, given that the 2003 and the 2007 CHIS show that health insurance, having a usual source of care and physician visits vary by legal status for Mexican-origin adults (Ortega et al. 2007; Vargas Bustamante et al. 2012). If barriers for adults have spillover effects, then parental legal status may be a factor in children’s health care. One study based on the 2001–2005 CHIS shows lower insurance coverage and health care utilization when the child and the surveyed parent are undocumented, as compared to when both are citizens (Stevens, West-Wright and Tsai 2010). This study did not identify Mexican-origin children per se or consider the immigration statuses of both parents as dyads.

Confounders and Mechanisms

Although parsing distinctions between confounders and mechanisms is difficult in cross-sectional research, legal status is associated with forms of capital that must be taken into account before concluding that it is a unique determinant of health care. The first consideration is financial constraints. The undocumented lack access to employer-sponsored health plans due to their concentration in secondary labor markets that offer low-wage jobs with few benefits (Bean et al. 2013). Paying for third-party insurance is also relatively difficult for them because they earn less than permanent residents, who earn less than the native born (Hall, Greenman and Farkas 2010).

To some extent, these differences may reflect endowments of human capital. Undocumented Mexican immigrants tend to have less education than permanent residents, naturalized citizens and native-born citizens (Bean et al. 2013). Although the likelihood of poverty is associated with levels and returns to schooling, education is also important for non-economic reasons. It affects health knowledge, awareness and efficaciousness. This is why children of parents without a high school degree are the least likely to see physicians, even after income and insurance are controlled (Yu et al. 2006; Ziol-Guest and Kalil 2012). Therefore, parental education must be accounted for in assessing status-related differences in health care access and utilization among children.

Language is another status-related barrier. The undocumented are less likely than naturalized citizens to be English proficient, in part due to the structure of incentives to acquire a second language (Bean et al. 2013). For instance, learning English may be less beneficial to undocumented residents who lack pathways to citizenship, good jobs and community integration. In turn, English mastery has implications for health care because it bolsters the capacity to navigate the health-care system; that is, to communicate with providers and to process information (Derose, Escarce and Lurie 2007; see also Vargas Bustamante et al. 2009; Vargas Bustamante et al. 2012; Kinchelow, Frates and Brown 2007; Lebrun 2012; Ortega et al. 2007; Yu et al. 2006).

The Conduciveness of California’s Economic and Social Climate

Most studies assume that the association between legal status and health care is time invariant. However, substantial changes have occurred in California during the past fifteen years that call this into question. Among the most far-reaching are those tied to the economy. Insecurity has fluctuated with economic shocks from a downturn in 2001 followed by the Great Recession from 2007 to 2009. The recession entailed long-term unemployment and the loss of over a million non-farm jobs (California Employment Development Department 2014). This was accompanied by a housing market collapse that resulted in the loss of substantial equity among homeowners; net worth fell among those who went “under water” with mortgage payments on property that depreciated in value.

Severe economic shocks would be expected to dampen health care access and utilization, but their implications for legal status differences are not straightforward. Downturns should exacerbate differences if the undocumented disproportionately experience budget constraints and are unable to rely on the public safety net as a buffer. Conversely, shocks should level differences if economic insecurity and scarcity are ubiquitous among those who are concentrated in “bad” jobs (even in the best of times). This implies greater reductions in health care by permanent residents and citizens, who may suffer disproportionately from losing “good” jobs in a deteriorating economy.

The conduciveness of the social environment must be acknowledged as well. Some scholars speculate that health care was a casualty of rising anti-immigrant sentiment. A “threat narrative” emphasizing immigration-related ruptures to the fabric of American society diffused through the media during the past decade (Chavez 2014). This narrative permeated the everyday lives of Mexicans, a group portrayed as overburdening the health care system (Rodriguez and Paredes 2014). In so doing, it undergirded efforts to identify and deport the undocumented. Hagan, Rodriguez and Castro (2011: 1378) claim that in North Carolina growing fears of deportation forced withdrawal from public settings as the undocumented “shied away from churches, schools and health services” (also, Berk and Schur 2001; Heyman 2014: 125). Rosenblum and Meissner (2014:41) argue that deportation had “ripple effects throughout…immigrant communities” and created “an atmosphere of fear and vulnerability extending to the broader community.” Others reported “recurring backlashes” against California’s immigrants (Schrag 2006). Such backlashes may spread to adjacent legal status groups and generate distrust, even among co-ethnics (Abrego 2014). In short, economic decline and an antagonistic social milieu would be expected to have negative repercussions for health care use among the undocumented. In turn, this would exacerbate status-based differences unless there were “ripple effects” that inhibited health care in other groups.

Another consideration is recent institutional reforms designed to facilitate access to health care. The 2010 Affordable Care Act (ACA) was implemented in the state under the banner of Covered California to increase insurance coverage by coordinating enrollment in state programs and plans offered by non-government entities through market-based exchanges. Covered California serves those who face financial constraints, but have incomes that exceed the maximum for Medi-Cal. Assistance is provided for insurance premiums and program participants are not charged deductibles or co-pays for preventive well-child care. This legislation also imposed tax penalties for not having insurance.³ Because it is intended for those who are “lawfully present” in California, undocumented parents can apply for their citizen children with the assurance that the application process is confidential and the receipt of assistance will not endanger the chances for naturalization. In short, pressure for insurance coverage started building in 2010 when the ACA was signed into law and accelerated with the development of community outreach and media campaigns before enrollment started in 2013. The ACA may have contributed to the widening of status-related differences due to changes in awareness and incentives for legal residents to purchase insurance.

Although excluding the undocumented from federal programs created under the ACA is a telling reminder of their marginalization, there are signs that receptivity for this group has not deteriorated. The emergence of children’s health initiatives was accompanied by modest employer compliance with the federal requirement to verify that employees are authorized to work (E-verify), the reluctance of many local police agencies to participate in federal programs that conferred authority to detain the undocumented, and active resistance to federal legislative proposals for restricting access to jobs and benefits (Leerkes, Leach and Bachmeier 2012).⁴ Moreover, public opinion has grown more tolerant. According to The Field Poll, from 1995–2006 the share of Californians who were “extremely” concerned about “illegal immigration” in the state declined from 58% to 43%. The percentage claiming that it had favorable consequences increased from 26% to 47% (Field Research Corporation 2006). The percent of registered voters approving a path to citizenship for the unauthorized increased from 75% in 2006 to 90% in 2013 (Field Research Corporation 2013). Such shifts may reflect changing views about the net benefit of undocumented migration. A recent USC/Los Angeles Times poll (2013) indicates that a majority of voters claim “illegal immigrants” have a positive impact on California’s economy (53% positive, 36% negative).

Public opinion about the government safety net is more complicated. Just one-fourth of registered voters agree that the undocumented should have the same health assistance as other state residents (Field Research Corporation 2012). Still, in 2007 56% favored guaranteed medical coverage for children in low-income families regardless of their immigration status (Baldassare 2007). In 2012, 44% opposed efforts to prohibit taxpayer-funded health care and education for children described as here “illegally” (USC/Los Angeles Times 2012). Such signs of increasing tolerance signal a more favorable climate for health care among the undocumented that could reduce status-related barriers.

Research Objectives

This research investigates health care access and utilization (hereafter “health care”) among Mexican-origin children in California from 2001 to 2014, a period marked by decline in the undocumented population in the aftermath of the recession (Passel, Cohn and Rohal 2014). The first objective is descriptive: to provide empirical evidence of the extent to which health care and the legal status composition of this population are changing. Legal status is important because it is typically hidden due to data limitations. The second objective is analytic: to investigate the implications of legal status for health care by identifying the particular distinctions that matter against the background of time. The analysis assesses whether differences by legal status are intensifying or dissipating, as well as changes in the salience of specific distinctions based on nativity, citizenship and documentation status. Needless to say, a multivariate approach is necessary to determine whether the relationship between health care and legal status is uniquely attributable to legal status per se or can be attributed to forms of capital it is associated with.

The descriptive portrait focuses on the legal statuses of both parents and children. It reveals substantial changes in California’s Mexican-origin population that prevent examination of the full array of children’s statuses. As a result, attention is redirected to their nativity. Child nativity is of interest both in its own right and as a variable that must be controlled to garner insights into the linkages between parental status and health care. The analytic framework accomplishes this with a two-pronged approach: by treating child nativity as a covariate in multivariate analyses of all children and by restricting the sample to those born in the United States. The former permits findings to be generalized to the total population while the latter resonates with studies devoted to “mixed-status” families of native-born children with undocumented parents (Fix and Zimmermann 2001; Yoshikawa 2011). In addition to these approaches, we provide information on health care access and utilization among children who are undocumented themselves. The additional steps that we take to bring these vulnerable children into view are described below.

DATA AND METHODS

This study uses restricted data from the California Health Interview Survey. The CHIS is a telephone survey that was conducted biennially from 2001–2009 and continuously from 2011–14. It provides information on representative samples of adults, adolescents and young children with separate questionnaires for each group. Our analysis is restricted to the “child” survey, which focuses on one randomly selected person aged 0–11 per household. This survey was administered to the parent in the household who was most knowledgeable about the focal child’s health.⁵ The total sample for our analysis consists of 21,441 children of “Mexican,” “Mexican American” or “Chicano” ancestry with parents in the United States.⁶

Dependent Variables

Access refers to the potential ability to secure services. Consistent with prior research, we focus on insurance coverage (Scott and Ni 2004; Prentice, Pebley and Sastry 2005; Stevens, West-Wright and Tsai 2010; Weathers, Minkovitz, Diener-West & O’Campo 2008). Children covered by public or private health insurance for the previous full year are contrasted with those who lacked coverage. This measures access to the resources of any third-party payer. The intricacies of access to specific forms of insurance and providers are tangential here, but important for future investigations.

Utilization is measured with physician visits. This dichotomous variable classifies children according to whether they saw “any kind of medical doctor” in the previous year. An annual examination is consistent with current guidelines of the American Academy of Pediatrics for children age 3+ without significant health problems (www.aap.org). Multiple visits per year are recommended for younger children, with timing and frequency determined by months of age. Given that the adequacy of the number of physician visits for young children cannot be determined without month-specific event histories for preventive care, we exclude those under three years of age in analyses of this variable. One visit per year among the remaining children complies with guidelines.⁷

Independent Variables

Parental Legal Status and Child Nativity Status

The immigration status of the mother, father and child can be determined from separate questions for each. These questions identify those who are native born, naturalized citizens, permanent residents and not permanent residents (termed “undocumented”).⁸ As shown in the cross-tabulation of mother’s status and father’s status in Table 1, an unwieldy 16 categories would be required to identify all possible combinations. Moreover, there are too few observations in some cells to support such detail. Instead, parent legal status is operationalized here with a modified version of a widely-used strategy that classifies children according to the mother or father with the lowest status (Oropesa, Landale and Hillemeier 2015; Ziol-Guest and Kalil 2012). Under conventional low-status anchoring, children with at least one undocumented parent are identified first. These children are shown in the largest backward “L” shaped boxed area at the perimeter of the table. Next, children with at least one permanent resident parent are identified. This is repeated for the remainder with at least one naturalized citizen parent, leaving two native-born parents as the last combination.

Table 1.

Coding Scheme for Parent Legal Status: Modified Low-status Anchoring Strategy (N=21,441)

Note: The percentage shown is for each unique combination of mother’s and father’s status, using weighted data from the 2001–2014 surveys. Boxed areas identify groups according to a low-status anchoring strategy. Shaded areas identify combinations that are preserved to identify children who have two parents with the same status.

We modified this approach by identifying parents with the same status. Although there are too few cases to do so for naturalized citizenship, children with two undocumented (21%, shaded cell “P”) and two permanent resident parents (8%, shaded cell “K”) are separated from those with one of each to arrive at a six-category classification. Due to widespread interest in those with the fewest resources, two undocumented parents is the primary reference group. This group stands apart from others, including those with one undocumented parent who may have greater access to resources through their permanent resident or citizen parent. This operational strategy indicates whether the number of parents with specific statuses matters. Supplementary information on contrasts using two native-born parents as the reference is also provided to ascertain how children of authorized immigrants (naturalized citizens, permanent residents) fare in comparison.

Some terminological leeway is required for efficient communication about these combinations. For example, reference to children with “one permanent resident parent” denotes “one permanent resident parent and one who is a naturalized or native-born citizen.” Children with “one permanent resident and one undocumented parent” are characterized as having “one undocumented parent.” Naturalized citizen dyads have one or two parents with this status, but none who are non-citizens.

The analysis of children’s legal status is complicated by sparse frequencies for some categories. Analyses of the total sample focus on child nativity instead to maximize inclusiveness. This sparseness is an essential part of the story that unfolds below. Nonetheless, insights into the largest segment of the foreign-born child population—undocumented youths with two undocumented parents—are provided.⁹

Covariates

Multivariate procedures incorporate forms of capital as controls, along with year of survey. Financial resources are measured by family income in relation to the federal poverty threshold. “Below poverty” is the reference category for a set of dummy variables that identify children in families with incomes that are 100–199%, 200–299% and 300+% of the threshold. Human capital is measured by the most knowledgeable adult’s education and English proficiency. High school graduates, those with some college and college graduates are compared to those without a high school degree (reference). Adults who speak English “only/very well/well” are contrasted with those who speak it “not well/not at all.” The most recent survey(s) is used as the reference for comparisons based on year.

Additional variables are family structure, child’s age and county of residence. Children of single parents are identified because family structure affects resources. Age is important if older children are seen as needing less routine care. County is controlled as an administrative unit for state and foundation-supported services. Using Los Angeles County as the reference, 13 counties with the largest concentrations of Mexican-origin children are separately identified along with a residual category for “all others.” The specific counties account for 83% of these children in California.¹⁰

The results for the control variables are not shown due to space constraints and research priorities. The key issue here is whether legal status matters independently of various characteristics that it is associated with. The associations between the covariates and statuses are shown in the appendix. Unsurprisingly, those at the bottom of the status hierarchy typically face greater disadvantages.

Methodological Issues

Several methodological issues should be noted. First is the response rate. Used by the state to monitor progress in achieving disease prevention objectives (http://www.cdph.ca.gov), the CHIS has not been insulated from the growing reluctance of the public to participate in telephone surveys, despite employing state-of-the-art data collection procedures. The response rate for the child module has fallen from 33% to 13% over time. However, confidence in these data is bolstered by the inability to infer bias from the level of non-response because the association between the two is weak (Groves 2006; Groves and Peytcheva 2008; Pew Research Center, 2012; Massey and Tourangeau, 2013). Some studies show that this type of bias does not threaten the representativeness of the CHIS (Lee et al. 2009). The response rates of this survey are also similar to those of the Behavioral Risk Factor Surveillance System (California Health Interview Survey 2014).

The second issue concerns trepidation about the measurement of legal status due to item non-response (National Research Council 2009). The possible reluctance of the undocumented to answer some questions is understandable given their interests in avoiding detection, but item non-response for queries about immigration status is not necessarily pronounced for non-governmental surveys. Regardless, profiles of this segment of the population are consistent across data sources that vary in the amount of incomplete information (Bachmeier, Van Hook and Bean 2014). Documentation status was imputed for fewer than five percent of Mexican-origin parents in the CHIS.

Another issue is the operationalization of parental legal status. The availability of information on both parents for all children is a potential advantage of this survey over others that just collect data on those in the same household (e.g. the SIPP). This additional information is valuable to the extent that non-coresident parents affect access to key resources, such as insurance, and are involved in the lives of their children. As a sensitivity test, we replicated the analysis by simulating a household-based survey that excluded absent parents. The results were not affected by this decision.

Fourth, omitted variables are a concern for all non-experimental studies. This is the case for the CHIS, a telephone survey designed to monitor the health of all Californians. It does not include information on pre-migration characteristics, the timing of transitions in legal status, the division of labor in families and numerous other characteristics that can be imagined. Still, information on legal status and key covariates permits analyses that compensate for the failure of other widely-used health surveys to include questions on the status of non-citizens (e.g. NHIS).

Last, the inability to determine whether care was received in Mexico or from providers who are outside the formal health care system does not have serious implications here. A 2001 CHIS supplement shows that virtually all native-born residents and the vast majority of immigrants in the Mexican-origin population use U.S.-based providers (Wallace, Mendez-Luck and Castañeda 2009). Just 3% of Mexican-descent adults sought care from a traditional folk healer (Lee et al. 2010).

Analysis Plan

The next section starts with a descriptive overview of trends in immigration status and health care, followed by multivariate logistic regression models. All parameter estimates are based on weighted data with standard errors adjusted for the complex sample design using SAS survey analysis procedures. The analytic framework for these analyses is expansive, with estimates presented separately for all Mexican-origin children and the native-born subset. Consistent with our primary interest in providing a comprehensive portrait, we focus disproportionately on the former. U.S.-born children are examined separately because of their increasing numeric dominance and interest in mixed-status families comprised of citizen children and undocumented parents (Fix and Zimmermann 2001; Landale, Thomas & Van Hook 2011; Vargas 2015, Yoshikawa 2011). As revealed below, the analysis also encompasses concerns about those who are potentially the most vulnerable—undocumented children.

RESULTS

Aggregate Trends

Table 2 shows the changing composition of Mexican-origin children with respect to their own and their parents’ statuses. Both distributions are associated with year of observation (Wald χ²’s). The top panel shows that the overwhelming majority of children were native born at every time point, but this group is increasingly prominent. It has grown from 92% to 98% over the past 15 years, presumably due to the impact of declining immigration on the undocumented population. Undocumented children have dropped from 6–7% to 1% of the total. This is noteworthy because the undocumented comprise the largest segment of the foreign-born population. About eight out of every ten foreign-born children were unauthorized prior to the recession. Since 2009, this figure has hovered near six out of ten.¹¹ Thus, the undocumented and the foreign born are dwindling in size.

Table 2.

Legal Status Distribution: Weighted Percentages for Mexican-Origin Children Ages 0–11

	Total Percent	Year
		2001	2003	2005	2007	2009	2011–12	2013–14
All Children
Child Status
US Born	94.8	92.3	91.5	93.9	94.3	96.5	96.7	97.5
Foreign Born	5.2	7.7	8.5	6.1	5.7	3.5	3.3	2.5
Naturalized Citizen	(0.5)	(0.7)	(0.4)	(0.3)	(0.7)	(0.6)	(0.6)	(0.3)
Permanent Resident	(0.9)	(1.1)	(1.4)	(0.7)	(0.7)	(0.7)	(0.6)	(0.8)
Undocumented	(3.8)	(5.8)	(6.7)	(5.1)	(4.3)	(2.2)	(2.1)	(1.4)
Total	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0
	Wald χ2 = 99.4, p < .001
Parent Status
Two US Born	29.7	25.9	24.5	29.2	30.2	27.8	32.7	36.0
Naturalized Citizen	13.3	12.4	12.2	14.1	13.9	12.8	12.7	14.6
One Permanent Resident	14.8	17.2	17.2	13.9	16.4	14.1	13.8	12.0
Two Permanent Residents	7.9	10.9	10.2	8.0	6.2	9.4	6.4	4.7
One Undocumented	13.7	13.7	14.2	13.7	12.2	14.8	13.2	14.3
Two Undocumented	20.6	19.9	21.7	21.0	21.1	21.0	21.1	18.4
Total	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0
	Wald χ2 = 110.2, p < .001
N0–11	21,441	4,066	2,739	3,567	3,186	2,928	2,879	2,076
U.S.-Born Children
Parent Status
Two US Born	31.3	28.1	26.7	31.1	32.0	28.8	33.8	36.9
Naturalized Citizen	13.8	13.2	13.2	15.0	14.7	12.9	12.8	15.0
One Permanent Resident	15.1	17.7	18.3	14.3	16.6	14.1	13.9	12.2
Two Permanent Residents	7.7	11.1	10.1	8.3	6.1	9.4	6.3	4.2
One Undocumented	13.7	13.6	14.5	13.8	12.1	14.8	13.4	14.0
Two Undocumented	18.2	16.2	17.2	17.5	18.5	20.0	19.8	17.8
Total	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0
	Wald χ2 =114.88, p < .001
N0–11	20,221	3,756	2,523	3,358	3,019	2,782	2,756	2,027

Note: All percentages are based on weighted data and N’s are based on unweighted data. The Wald log-linear χ² test for odds ratios is reported, with p-values based on an adjustment for the degrees of freedom (also known as Wald F_adjusted). Inferences from these tests are identical to those for design-corrected Rao-Scott χ² tests.

In contrast to children, there is substantial variation in parental statuses in the top panel. Half of all children currently (2013–14) have mothers and fathers who are citizens; 36% have two native-born parents and 15% have at least one naturalized parent. Another 12% are the offspring of one citizen and one permanent resident. This leaves 5% with two permanent resident parents and a third with one (14%) or two (18%) who are undocumented. However, temporal change is substantial (Wald χ² = 110.2, p<.001). The current share with two native-born parents is the culmination of an upward trend in the size of this group from one-fourth to over one-third of all children. This reflects declines in those with a green card, and perhaps the undocumented in the last survey.

As stated above, the distribution of children’s legal status has implications for the analysis. Because a small and declining share is foreign born, children’s status cannot be examined in detail. Instead, nativity is emphasized with the understanding that a large majority of the foreign born are undocumented. The analysis focuses primarily on the total sample of Mexican-origin children for insights into the consequences of nativity (children) and legal status (parents) for health care. Analyses restricted to U.S.-born children are also provided for additional insights even though inferences for the native born and all children are generally identical.

Declines in immigration and the economy have different implications for aggregate trends in health care (ceteris paribus).¹² The loss of newcomers who are disproportionately marginal to the health care system would be expected to exert upward pressure on trends with the growth of the native-born population. Downward pressure on trends would be expected from the intensification of financial constraints and the loss of employment benefits during a period of economic decline.

Table 3 shows no evidence of deterioration. The likelihood of insurance coverage is significantly associated with year for all children (Wald χ² = 90.6, p<.001). The percent insured rose from 81% in 2001 to 87% in 2003 before inching up to 93% in 2013–14. As indicated by the asterisk notation, coverage in the most recent year is higher than that for each prior year except 2011–2012. This pattern is replicated for the native born in the bottom panel. Improved access is unmistakable.

Table 3.

Dependent Variables: Weighted Percentages for Mexican-Origin Children

	All Years	Year
		2001	2003	2005	2007	2009	2011–12	2013–14
All Children
Insurance
% Insured	88.3	81.0***	87.1***	86.6***	87.4***	89.8**	91.3	93.2
N0–11	(21,441)	(4,066)	(2,739)	(3,567)	(3,186)	(2,928)	(2,879)	(2,076)
	Wald χ2= 90.6 p < .001
Physician
% Visit	88.5	86.1**	88.0	88.9	87.3*	88.5	89.5	90.8
N3–11	(16,137)	(3,043)	(2,082)	(2,514)	(2,330)	(2,222)	(2,242)	(1,704)
	Wald χ2= 11.1, p < .10
U.S.-Born Children
Insurance
% Insured	89.8	83.4***	90.5*	88.0***	89.3**	90.2*	92.0	94.1
N0–11	(20,221)	(3,756)	(2,523)	(3,358)	(3,019)	(2,782)	(2,756)	(2,027)
	Wald χ2=72.1, p < .001
Physician
% Visit	89.4	87.8*	89.4	89.3	88.1+	89.4	90.0	91.2
N3–11	(14,974)	(2,748)	(1,871)	(2,320)	(2,171)	(2,082)	(2,125)	(1,657)
	Wald χ2= 6.9, p = n.s.

Note: Percentages are based on weighted data, with unweighted N’s for children ages 0–11 (insurance) and 3–11 (physician). The Wald log-linear χ² test for odds ratios is reported, with p-values based on an adjustment for the degrees of freedom (i.e. to transform the test statistic into Wald F_adjusted). Inferences from these tests are identical to those for design-corrected Rao-Scott χ² tests. Asterisk notation is used to denote the significant differences between 2013–14 and each prior year: + p <.10, * p<.05, ** p<.01, *** p<.001.

It should be noted that these figures reflect a change in type of insurance. Half of all Mexican-origin children were insured by public programs from 2003–2007. After jumping to 64% in 2009, this share has grown to two-thirds due to declining employer coverage. Those covered under insurance provided by an employer fell from 37% prior to the recession to below 30% over the past five years. Thus, the public safety net has increased in importance.

As for health care utilization, the percent seeing a physician is stable for all children and the native born despite expanding access to insurance. Values range from 86% to 91% in the top panel and 88% to 91% in the bottom panel, with p-values of the test statistics for both failing to cross the .05 significance threshold (e.g., Wald χ² = 11.1, p=.09 for the total sample). There are only two years (2001 and 2007) in which physician visits differed significantly from the value of 91% for 2013–14. The small magnitude of differences testifies to the lack of substantial variation by year.

Health Care and Status

The second objective was to elucidate the associations between health care and the statuses of parents and children against the backdrop of time. This requires attention to temporal patterns in the presence, strength and category-specific nature of associations.

Insurance

Table 4 presents the percent insured by parental immigration status and children’s nativity status for each year to facilitate two types of comparisons. First is the effect of year within status, which tells us how insurance coverage has changed over time for specific groups of children. Tests for the significance of year within each specific status category are reported in the last column (Wald χ²). Letter superscripts identify estimates within each row that are significantly different from the value for 2013–14. Second is the effect of status within year. The last row for parental status presents the statistical test for the overall association (Wald χ²), with asterisk notation identifying significant differences between each specific category and two undocumented parents within each year.¹³ “Borderline” significance, defined as p<.10, is indicated to safeguard against a Type II error due to the diminished statistical power of tests for smaller groups. Lastly, the consequences of switching the reference category to those with two U.S.-born parents are shown. Shaded cells that are outlined identify significant contrasts at p<.05 and shaded cells without an outline are for p<.10. Thus, the shaded value of 85.8 for one undocumented parent indicates that it is significantly lower than the value of 92.9 for two native-born parents (Panel A, Column 1).

Table 4.

Insurance: Percent Insured by Status and Year for Mexican-origin Children Ages 0–11

Note: Within years, asterisks identify significant differences using two undocumented parents as the reference for parental status and the U.S. born as the reference for child’s nativity: + p <.10, * p < .05, ** p < .01, ***p < .001. Significant contrasts with two U.S.-born parents as the reference are shown with outlined shaded cells for p<.05 and unlined shaded cells for p<.10. For within-status comparisons across time, 2013–14 is the reference and significant differences with prior years are indicated by letter superscripts: a, p<.10; b, p< .05; c, p<.01, d, p<.001. Overall tests for associations are based on Wald’s χ² log-linear tests for odds ratios. In Panel B, this test is based on U.S.-born children only. The asterisk notation for undocumented children with two undocumented parents identifies significant difference between them and US-born children of two undocumented parents. These values were generated with a separate analysis.

These results reveal a major transformation in access to insurance that is marked by the dissipation of differences across categories of parental status due to increases in coverage among those at the lower end of the hierarchy. The pattern for the total sample with all years combined suggests modest differences that are rooted in citizenship (column 1). 93% of children with two native-born parents and 90% of children with two naturalized-citizen parents were insured, with the former significantly greater than the values for those with permanent resident parents (82%), one undocumented parent (86%), and two undocumented parents (85%). Contrasts between both of the permanent resident categories and two undocumented parents approach significance as well due to the large sample (p<.10), but gaps of 2–3 percentage points are of little practical importance.

While informative, these results obscure the “historical” record. The record shows that an essentially citizen-noncitizen distinction in the first few years of the decade dissipated around the same time as the recession and subsequent decline in immigration. This can be traced to increasing coverage among those with undocumented parents, not from decreasing coverage among the native born. Focusing first on the effect of year within each status, the last column shows that the likelihood of being insured is not associated with year for those with native-born and naturalized citizen parents. However, insurance is increasingly associated with year as one moves down the status hierarchy toward two undocumented parents (Wald χ² = 52.6, p<001). Among children with one and two undocumented parents, coverage rose dramatically from 74% in 2001 to almost 95% in 2013–14. Improvement is also evident for those with one parent who has a green card. The pattern is less regular for those with two parents who have green cards.

This provides the foundation for viewing temporal changes in status-based differences within each year. The test statistic in the last row for parental status shows that it has been inconsequential since 2009 for all children and the native born, with parity evident across most categories. In 2001, there were large citizenship-related differences: 90% of children with two U.S.-born parents and 86% with naturalized parents had insurance (83% for permanent resident and citizen parents). Children of two permanent residents, one undocumented, and two undocumented parents were all at 73–74% (χ² =79.6, p<.001). In 2013–14, more than nine of every ten children in the top three categories had insurance, but this was also true of children with one or two undocumented parents (χ² =8.9, p = n.s.). Children with two permanent resident parents are the only group to stand out, with 80% insured.¹⁴

Insurance is strongly associated with child nativity, with 90% of the native born and 60% of the foreign born covered across all years combined. However, this 30 percentage-point nativity gap is misleading in view of trends. The share of native-born children with insurance grew unevenly from 84% in 2001 to its current high of 94%. Among the foreign born, insurance coverage rose from 50% to a high of 80% in 2009. This reduced the nativity gap to 10 percentage points. Since then, the rate for this group has returned to earlier levels (60%).

Panel B provides results for parental status among native-born children that largely replicate those for the total sample, except for reductions in some differences. To avoid redundancy, we focus on two groups of children with undocumented parents who are of special interest: the native born and the undocumented. The values for undocumented children of undocumented parents (shown in parentheses) were generated with additional analyses that added this group to the native-born sample. These values indicate that differences are narrowing.¹⁵ About 50% of undocumented and 80–90% of U.S.-born children of undocumented parents had insurance prior to 2009 (except 2005). U.S. birth confers an advantage for these children, but the gap has become less pronounced with coverage among undocumented children rising to 74–80% for two of the last three surveys.

Table 5 presents odds ratios from logistic regression models. Year-specific models pool adjacent years in order to increase the stability of estimates and decrease the size of standard errors. Using the same notation as before to denote significant contrasts, the columns on the left show results for all Mexican-origin children and those on the right are for the native born. Model 1 for both reveals differences in insurance by parental legal status, controlling for child nativity. Child nativity is controlled by including it as a predictor in analyses based on the full sample and by restricting the sample to U.S.-born children. Model 2 includes the full set of covariates.

Table 5.

Insurance: Multivariate Odds Ratios from Logistic Regressions (N=21,441)

Note: Within years, asterisks identify significant differences using two undocumented parents as the reference for parental status and the U.S. born as the reference for child’s nativity: + p <.10, * p < .05, ** p < .01, ***p < .001. Significant contrasts with two U.S.-born parents as the reference are also shown with outlined shaded cells for p<.05 and unlined shaded cells for p<.10. For within-status comparisons across time, 2013–14 is the reference and significant differences with prior years are indicated by letter superscripts: a, p<.10; b, p< .05; c, p<.01, d, p<.001. Asterisk notation for undocumented children with two undocumented parents identifies significant difference between them and US-born children of two undocumented parents. These values were generated with a separate analysis. Model 2 includes year as a covariate for pooled surveys.

A striking feature of Model 1 for the full sample is that few differences between children with two undocumented parents and other groups are significant, once child nativity is controlled. This suggests that the likelihood of having insurance is associated with parental status primarily because of child nativity, especially from 2003 onward.¹⁶ At the same time, another pattern is evident when children with two U.S.-born parents are the reference. Despite several anomalies, these results generally indicate that children with two native-born parents have a higher likelihood of insurance coverage than other children with the exception of those with two naturalized citizen parents.¹⁷

Model 2 for the full sample shows that most of the remaining differences are explained by the additional controls. The F statistics (Type 3 effects) indicate that insurance coverage is associated with parental legal status only in the model in which all survey years are combined. Furthermore, it is important to note that there are few significant differences in the most recent period either before (Panel A) or after the covariates are controlled (Panel B). Currently, parental status is not a unique source of variation in coverage. In contrast, estimates for child nativity are impervious to model specification. The likelihood of being insured is negatively and strongly associated with foreign birth. Foreign-born children are considerably less likely than the native born to be covered, regardless of parental status, year and other factors.

The parallel analysis for U.S.-born children on the right side of the table shows results for Model 1 that are similar to those for the full sample, particularly the consistent difference between children with two U.S.-born parents and others. Those for Model 2 (ignoring the odds ratios in parentheses) also mirror those for the full sample. A key finding is that native-born children of two undocumented immigrants are not uniquely disadvantaged in terms of the likelihood of being uninsured.¹⁸

Odds ratios in parentheses are the entry point for returning to the question: Does the likelihood of insurance coverage differ for the U.S.-born (the reference) and the undocumented children of two undocumented parents? Based on a separate analysis, these estimates demonstrate a persistent advantage of being born in the United States for those with undocumented parents. Model 1 shows that relative to their native-born counterparts, undocumented children with undocumented parents are extremely unlikely to be covered. In 2011–14, the odds for the undocumented are one-tenth the odds of the native born with two undocumented parents (.12, p<.001). Values for other years in the .16–.19 range convey substantial disadvantage as well.

Physician Visits

Table 6 displays bivariate associations that tentatively point to the re-emergence of citizenship rather than documentation status as a basis for differentiation in health care utilization. Focusing first on all children, the test statistics (Wald χ²) indicate that parental status mattered in the earliest and most recent years. From 2011–14, about 94% of children with two U.S.-born parents, 91% with at least one naturalized parent and 92% with one citizen and one permanent resident parent saw a doctor annually. These percentages tend to be similar to one another and higher than those for children with two undocumented parents (83–87%). In turn, the percentages for those with one undocumented (90%) and two permanent resident parents (81%) are not significantly different from those for children with two undocumented parents. It is worth noting that recent results for those whose parents have green cards are pivotal for conclusions about citizenship versus authorization. The values for children of two permanent residents are consistently among the lowest observed and frequently lower than the values for children of two native-born parents, even though only one of the 13 percentage-point differences between these groups for the last two surveys is significant (94% vs. 81%, 2011–12).

Table 6.

Physician Visits: Percent Seeing a Medical Doctor by Status and Year for Mexican-origin Children Ages 3–11.

Note: See Table 4 for a description of the notation.

Such differences were more strongly evident in 2001 before subsequently dissipating. In the first survey, health care utilization among children of two native-born parents and children of two naturalized citizen parents tended to be similar. These groups also stood apart from others. However, children with one permanent resident and one naturalized parent have become indistinguishable from those with two citizen parents since 2001. The tests statistics in the last column indicate that this is the only category for which utilization is significantly associated with year, but the pattern is neither linear nor curvilinear (i.e., idiosyncratic values for 2001 and 2007 are responsible). As with the aggregate trend, temporal stability predominates; it characterizes the rates for five of six groups even though some year-to-year fluctuations exist.

Regarding child nativity, the foreign born are considerably less likely than the native born to have seen a doctor at every time point except one. In 2013–14, 91% of the former and 78% of the latter saw a doctor. There is little evidence of meaningful change for the native born, with non-significant fluctuations of ±3 percentage points over time. In contrast, the pattern for the foreign born is intriguing even though the likelihood of seeing a doctor is unrelated to year. An initial nativity gap of seventeen percentage points (88–71) temporarily closed to a four percentage-point difference (89–85) by 2005 that is indistinguishable from zero due to increased utilization among the foreign born. The gap subsequently widened by 2009 before narrowing again most recently for the same reason as before. The earlier shift is consistent with what might be expected with the growth of Children’s Health Initiatives in the state while the recent shift is consistent with what might be expected from changes in immigration, economic decline and economic recovery.

Numerous results for native-born children in Panel B roughly mirror these, especially for recent years due to reductions in the size of the foreign-born child population. The impact of excluding the foreign born is largely confined to earlier years where differences between the extreme categories are eliminated. From 2001 to 2009, there was no difference between those at the bottom and top of the status hierarchy. In the last five years a significant difference in the likelihood of seeing a physician emerged between native-born children of two native-born parents and native-born children of two undocumented parents. This is evident for the total sample as well.

The last remaining contrast involves differences by nativity for those with two undocumented parents. The last two rows of percentages indicate that native-born children of undocumented migrants are substantially more likely than their undocumented counterparts (in parentheses) to have seen a physician at every time point except one. Moreover, the rate for undocumented children fell to around 65% from 2009 onward. In 2013–14, 89% of native-born children and 66% of undocumented children with two undocumented parents saw a physician. In sum, the child’s status remains important as a source of differentiation even among those whose parents are undocumented.

Table 7 presents odds ratios from logistic regressions, using the same format as before except for the addition of insurance coverage as a predictor in the full multivariate model.¹⁹ Models for all children tentatively point to the growing importance of parental citizenship to children’s health care utilization, despite the attenuation of some differences after the covariates are taken into account. Specifically, children in the top three categories of parental status have two parents who are authorized to live in the United States and at least one who is a citizen. Model 1 for 2011–14 shows that children in two of these categories—both parents native born (OR=2.8) or one citizen and one permanent resident parent (OR=2.0)—are more likely than those with two undocumented parents to have seen a doctor. In addition, all groups of children with no citizen parents are less likely than those with two citizen parents to have done so. This pattern is largely replicated in Model 2. Results for the most recent period also show that parental documentation status is not a key factor; none of the groups with no citizen parents differ significantly from children with two undocumented parents. Finally, the results indicate that child nativity generally matters for physician visits, with foreign-born children less likely to see a doctor yearly than native-born children.²⁰

Table 7.

Physician Visits: Multivariate Odds Ratios from Logistic Regressions (N=16,137)

Note. See Table 5 for a description of the notation.

For the U.S.-born sample, evidence remains of an emerging divide by parental citizenship in 2011–14, despite weaker associations overall.²¹ Native-born children whose parents were unauthorized residents had a substantially lower likelihood of seeing a doctor than two groups with citizen parents. The odds for children of the native born and those with one citizen and one permanent resident parent were both greater than the odds for children of the undocumented by ratios of 3:1 and 2:1 in Model 2, respectively. These odds ratios suggest an increasingly precarious situation for native-born children with undocumented parents, but these children remain more likely than undocumented children with undocumented parents to see a doctor (parameter estimates in parentheses in the last row of each panel). This pattern holds across all years in Model 1 and the most recent years in Model 2 (plus 2001). Currently, undocumented children are just half as likely as their native-born counterparts to see a doctor (.45, p<.10) after all covariates are controlled.

The preliminary analysis also explored the extent to which insurance is a mechanism for other associations. This involved 171 comparisons across categories for both measures of status. Eliminating insurance from the multivariate model affected the inferences for just 5% of the comparisons, perhaps what would have been expected by chance. The most noteworthy instance of this occurs for children with at least one naturalized parent in the pooled sample. The non-significant odds ratio of 1.64 for the 2011–14 sample changes to 1.90 (p<.10) when insurance is excluded. This strengthens the case for the emergence of a divide by citizenship.

A final issue is the anomalous findings for the 2007–09 period. For both all children and U.S.-born children, Model 2 suggests that those with two undocumented parents are more likely than those in all other groups except one (children with two permanent resident parents) to have seen a physician once various covariates are taken into account. In supplemental analyses we explored this in considerable detail to identify the mechanism that is responsible for this unexpected pattern. With standard errors that are stable across various alternative models, the analyses identified education as the key factor. Once their greater likelihood of having parents with low formal education is taken into account, children of undocumented parents are the most likely to have seen a physician. No other covariate could be identified as a source for this association.

It is difficult to determine why this emerged and did not extend beyond these years. The 2007–09 period was characterized by the reduction of migrants to the state with the onset of the recession, as well as efforts to extend the reach of Children’s Health Initiatives. In fact, 31 counties were participating in this effort to reach those who were ineligible for mainstream programs. By 2007, every county-based health initiative that would ever be underway was up and running. Nearly a third of these programs ceased operations by 2014.

CONCLUSION

The growth of the Mexican-origin population draws attention to the next generation in California. Many young residents of the state are personally connected to the immigrant experience as children of immigrants or as immigrants themselves. Yet, describing “the” immigrant experience is elusive for several reasons. Perhaps the most obvious is that legal status is a source of heterogeneity in life chances that is rarely measured. This determinant of inequality in access to opportunities and resources is especially important for children because they are dependents. Their life chances are potentially linked to both their own legal status and the statuses of their parents.

In keeping with their prominence as public issues, the first research objective was to provide an overview of trends in immigration status and health care among Mexican-origin children. The CHIS reveals dramatic changes in the legal status distributions of children and parents that have potential implications for health care. The native born increased from 92% to 98% of all Mexican-origin children over the past 15 years due to a decline in the undocumented. Undocumented children fell from 7% to just 1% of the total. Moreover, there has been a corresponding upward shift in the distribution of parental status. Children with two citizen parents grew from 38% in 2001 to 51% in 2013–14, primarily from the expansion of those with two native-born parents. Of course, the number of children with undocumented parents remains substantial. Over a third have one (12–15%) or two (18–22%) undocumented parents in every year. Such results imply that parent’s status should not be ignored in efforts to monitor inequality in health care access and utilization.

As to health care access, there was an upward trend in insurance coverage among Mexican-origin children. The percent insured increased from 81% to 93%, with most change confined to the earliest and latest years. This is intriguing given the expectation that coverage should mirror the economy; it is hard to see traces of a major recession in the temporal pattern. The percent of children who were insured increased after the recession of 2001, but the Great Recession (2007–09) did not leave a visible imprint on an aggregate trend that continued to inch upwards when a decline would be expected. It is also difficult to discern the state of various programs from the temporal pattern even though the rise in coverage after the early recession coincides with the emergence of Children’s Health Initiatives in 2001. The rate stabilized when initiatives gained momentum as they spread across the state before contracting with the downturn. Interestingly, coverage increased when programs started to lose momentum. This could reflect the influence of declining in-migration and increasing out-migration on the number of children who are most likely to lack insurance.

These changes appear to have had few aggregate-level consequences for health care utilization. The percentage of children seeing a physician has been stable at around 90% per year. This is also surprising given the expectation of declining utilization with severe economic shocks and increasing utilization with greater access to insurance, initiatives for hard-to-reach populations, and reductions in recent migrants who are unlikely to seek care. The extent to which these factors exerted countervailing pressures warrants additional scrutiny in future research.

The second objective was to investigate the connection between health care and legal status against the backdrop of time; that is, to show how the latter is changing as a source of differentiation. Locating the status-related boundaries that are critical to health care access and utilization required a comprehensive analysis of the full range of categories that constitute the status hierarchy. The conventional wisdom is that distinctions have hardened around documentation status (Coutin 2011; Marrow and Joseph 2015). In other words, unauthorized immigrants and the children of unauthorized immigrants are increasingly marginalized.

Basic descriptive statistics provided an indispensable vantage point for viewing the temporal pattern of differences, given that covariates are not controlled among those who need services in the “real world.” The most useful place to start is with the current state of affairs. Consistent with expectations, undocumented children are much more likely than native-born children to lack insurance (21% vs. 6%). However, future gains for Mexican-origin children as a whole may come from the 165,000 native-born children who are now uninsured, despite their greater likelihood of coverage (165,000 2,748,170_{Weighted N}*.06_{prop. uninsured}). This group dwarfs the 8,300 undocumented children now without insurance due to the fact that nearly all Mexican-origin children were born in the United States (8,300 39,657 *.21). Because the ratio of uninsured U.S.-born to undocumented children is 20:1, the native born will be the source of future gains in health care for the Mexican-origin child population as a whole unless immigration increases or universal coverage is unattainable. A similar story could be told for physician services.

The analysis also shows that nuance is necessary to address whether parental legal status has increased in salience as a source of inequality in health care. Status-related gaps in the likelihood of being insured closed over time due to substantial increases in coverage among children with undocumented and permanent resident parents. Children of non-citizens and citizens are increasingly indistinguishable. Moreover, differences by child nativity narrowed due to changes for the foreign born in general and the undocumented in particular. Fewer than half of undocumented children had insurance in 2001–2003. From 2009 onward, this grew to 70–80%. Change did not result from a decline among citizens in the aftermath of a recession.

A modest citizen-noncitizen divide was evident for health care utilization during some years, but increasing insurance coverage has not eradicated differences. In 2001–03, about 90% of children with two citizen parents saw a doctor, compared to 80–85% of those with undocumented or two permanent resident parents. In 2013–14, the figures for children of two U.S.-born parents, two citizen parents and one citizen parent were 91–95%. Those for two permanent resident or two undocumented parents were 81% and 87%, respectively. In addition, multivariate results for 2011–14 showed that the citizen-noncitizen distinction remains after all other covariates are controlled. Those with two native-born parents and one naturalized parent are twice as likely as those with two undocumented parents to see a physician. Children with two parents who have a green card, in turn, are similar to those with two undocumented parents. Such findings point to the emergence of citizenship as a source of differentiation. For children, the nativity gap narrowed before widening due to declines in utilization among the foreign born and undocumented. The salience of child nativity continues to reflect the salience of immigration.

Consistent with the emerging role of citizenship in health care, attention is required to the vulnerability of children of permanent residents. Mexican-origin children who have two parents with green cards are as vulnerable as their undocumented counterparts. Regardless of the sample examined, the offspring of two permanent residents have not been more likely than those in undocumented families either to have insurance or to see a physician over the past 15 years. In fact, native-born children of two permanent residents had the lowest levels for some years.

This finding is counterintuitive. Permanent residents do not face formal legal obstacles to employment or participation in public insurance programs (except for the wait-time rule in earlier years). However, these parents may be discouraged by a misunderstanding of eligibility or fear that program participation will disqualify them from naturalizing at a later date given prohibitions against becoming a public charge. Permanent residents are also disadvantaged in terms of access to resources, even though they are less likely than the undocumented to be impoverished. Children of two permanent residents are among those who are least likely to have an English-speaking parent who graduated from high school or college. Children of permanent residents also tend to be older, and older children may be seen as needing less routine care when resources are scarce. Thus, the consequences of having two permanent resident parents must be understood in the context of knowledge, beliefs, endowments of capital and age. These sources of vulnerability may contribute to the citizen-noncitizen divide in health care.

Lastly, the results for the native born put the circumstances of children in mixed-status families in sharp relief. Native-born children of undocumented migrants are often singled out as having a high risk of being uninsured and underutilizing the health care system because of the policy-related constraints their parents face. The descriptive results show that these children were not consistently vulnerable, even though such concerns have traction for insurance at the earliest time points if those with native-born parents are the reference. In contrast, citizen children of the undocumented did not have especially low chances of insurance coverage or doctor visits if compared to children of permanent residents. Importantly, nativity also matters: native-born children of undocumented parents are less marginalized with respect to health care than undocumented children.

This study also has implications for efforts to monitor trends. Most widely-used surveys such as the National Health Interview Survey fail to include questions on the status of non-citizens. The findings reported here indicate that inferences from surveys that are limited to information on nativity and citizenship may not be egregiously misleading. This tentatively supports analyses of datasets that lack information on legal status, despite uncertainty about whether low levels of immigration will continue and whether these findings can be generalized to other places.

Investigating how the legal statuses of parents may jointly influence health care is complicated by the absence of consensus on the appropriate strategies for combining information (Oropesa, Landale and Hillemeier 2015). Consistent with some prior studies, we classified children according to the mother or father with the lowest status. The key justification for this approach is that those with a parent(s) on the lowest rungs of the ladder have diminished opportunities. An alternative measurement approach could be grounded in the parent with the highest status, given that families are adaptable and strategic about using resources. Comparing such approaches is an avenue for future research, especially in conjunction with attention to the family as a gendered institution with a division of labor for domestic and market-based activities that may have implications for health care. Gender-blind approaches to understanding the influence of parental legal status should be examined closely to determine how responsibilities for children’s health care are organized.

The generalizability of these findings to other states is another important task for future research. Recent data from the American Community Survey (2010–12) suggest that 8.3% of U.S.-born Mexican-origin children in California do not have health insurance---a value near the median for all states (9.2%). This does not mean that California is typical of all states, states along the border, or states with large immigrant populations. On the contrary, California is one of two states with favorable state legislative activity for “irregular immigrants” (CONAPO 2010). Future studies must determine how state policy environments, population composition and economic structures affect status-related disparities in health care. States play a potentially crucial role in health care. Other administrative entities such as counties and municipalities are also potentially important (Marrow and Joseph 2015).

Related concerns surround the implementation of the Affordable Care Act under the banner of Covered California. The ACA was designed to increase access to health care by requiring residents to purchase insurance, with provisions to lower costs for those with the least ability to pay. A proposal is now before the state legislature to extend Covered California to the undocumented and the governor signed legislation (SB 75) in 2015–16 to fund the extension of full-scope Medi-Cal to children under the age of 19 whose legal status would make them ineligible for coverage (leginfo.legislature.ca.gov). It remains to be seen whether such reforms will be transformative.

Despite the risk of belaboring the obvious, uncertainty about the future remains because change is difficult to anticipate. Few anticipated the transformations that occurred in the Mexican-origin child population over the past sixteen years. Still, the dramatic growth of the native-born share is not a reason to turn a blind eye toward the undocumented. Apprehensions by the Border Patrol have increased, even though the annual total (421,000 in 2013) pales in comparison to more than a million apprehensions per year from 1983–2006 and just 10% of apprehensions occur in California (Simanski 2014). Nonetheless, monitoring these broader trends and their changing implications for differences in health care must remain important priorities.

Appendix.

Descriptive Statistics by Parent and Child Status: All Years (N=21,441)

	Below Poverty %	< High School %	English %	Single Parent %	Child Age Mean
Parent Status
Two US Born	20.7	9.9	99.2	33.2	5.1
Naturalized Citizen	19.1	18.1	83.3	21.6	6.0
One Permanent Resident	27.2	36.8	60.7	20.1	5.7
Two Permanent Residents	43.6	65.6	25.6	14.1	6.4
One Undocumented	54.6	53.7	41.9	34.9	5.0
Two Undocumented	69.3	67.7	10.1	25.8	5.1
Child Status
US Born	36.4	35.7	61.9	27.6	5.3
Foreign Born	65.9	66	13.1	15.0	7.7)
Naturalized Citizen	(38.7)	(49.2)	(44.2)	(24.0)	(7.7)
Permanent Resident	(59.1)	(71.4)	(15.4)	(5.8)	(7.8)
Undocumented	(70.9)	(67.0)	(8.7)	(15.9)	(7.7)

Note: Estimates are based on weighted data for 2001–2014. All associations are statistically significant at p < .001.