adulthood. The key difference between the 1.5 and 2 nd generations is citizenship: the 2 nd generation are citizens by birth. Because the 1.5 generation grows up and forms primary attachments in the country of destination, for 1.5-generation noncitizen immigrants, the exclusions of immigration law and threat of immigration enforcement create an especially painful and alienating experience that has consequences on schooling and work, connections with others, and identity (Abrego 2011;Gleeson and Gonzales 2012;Gonzales 2011;Patler 2018b;Vaquera, Aranda, and Sousa-Rodriguez 2017). The experience of growing up without status likely has unique consequences for health, as well, but research on legal status disparities in health has heretofore grouped the 1 st and the 1.5 generations.

In this study, we examine legal status disparities in health within and between 1 stgeneration and 1.5-generation adult immigrants of Latinx origins living in California. 1 Specifically, we compare the health of naturalized U.S. citizens, Lawful Permanent Residents (LPRs), and non-citizen, non-LPRs, who are likely to be undocumented. 2 We use data from Latinx immigrant adults interviewed in the 2005-2017 waves of the California Health Interview Study (CHIS), which provides a representative sample of Latinx immigrant adults living in California during this period. Because we are interested in broad patterns of health inequality, rather than specific causes of illness, we examine patterns across nine self-reported health outcomes measured in the CHIS, including chronic conditions, current conditions, self-health assessment, and mental distress. 1 We use the term Latinx to refer to people who were born in a country in Latin America. We opt to use the term "Latinx" rather than "Latino" or "Latino/a" to maintain gender neutrality. 2 As described below, an estimated 88-93% of Latinx noncitizen immigrants in California are undocumented (Vargas Bustamante et al. 2012). For simplicity, heretofore we refer to this group as "undocumented."

Immigrant rights range across what Joseph (2020) calls the "documentation status continuum" (see also Bean et al. 2015;Oropesa, Landale, and Hillemeier 2015;Patler 2018a;Waters, Pineau, and National Academies 2015). The rights granted by immigrant legal status dictate how long someone can be in the United States, the ability to transition into other (more permanent) legal statuses, the right to sponsor family for immigration, the right to work or change employers, the right to travel in and out of the United States, and the right to access federally funded services, such as health insurance, welfare benefits, and student loans. Among immigrants, those with the greatest rights are naturalized citizens, who through naturalization obtain the same rights as U.S.-born citizens. LPRs (also called "green card" holders) can remain in the United States indefinitely, have the right to work for most U.S. employers, and can sponsor some family members for immigration, but they cannot vote and are subject to deportation if convicted of certain crimes. Individuals with noncitizen, non-LPR status have still fewer rights. This group includes temporary visa holders (e.g. students, temporary workers, and refugees for the first year of residence) whose right to work and length of status depend on the terms of the visa, and those with "liminal" (Menjivar 2006) statuses such as Temporary Protected Status (TPS) or Deferred Action for Childhood Arrivals (DACA) status.

Undocumented immigrants, who entered without inspection or overstayed a temporary visa, have the fewest rights. Most significantly, undocumented immigrants do not have the right to work and are subject to removal on the basis of their immigration status.

The rights granted by one's position on the documentation status continuum determine three key mechanisms for health: socioeconomic status and resources, access to health care, and stress. Immigrant legal status is associated with socioeconomic disadvantage, especially between undocumented and other immigrants, as a result of legal barriers to the U.S. labor market, postsecondary education, and social safety net programs designed to keep people out of poverty (Amuedo-Dorantes and Antman 2016;Borjas 2011;Gleeson 2010;Phillips and Massey 1999).

As a result of barriers to legal work, as well as federal laws that bar non-citizen immigrants from Medicaid and other federally-funded insurance programs, undocumented immigrants have far lower rates of insurance coverage and access to health care (Gomez Cervantes and Menjívar 2020;Marshall et al. 2005;Vargas Bustamante et al. 2012). Not only does non-citizen legal status present formal barriers to socioeconomic opportunities and support, but surveillance of immigrants and enforcement of immigration law by local and federal law enforcement creates substantial stress and fear in immigrant communities (Asad 2020;Berk and Schur 2001;Dreby 2015;Enriquez, Hernandez, and Ro 2018;García 2018;Zayas et al. 2015). Immigration status therefore affects health through legal status-stratified systems of resources and opportunities, including health care, and the stress of immigration enforcement.

Because socioeconomic status, health care, and stress are key mechanisms linking social conditions to health, scholars have argued that immigration policy, or immigration status (itself determined by policy), is a social determinant of health (Asad and Clair 2018;Castañeda et al. 2015;Martinez et al. 2015;Taylor 2020). A growing body of research examines whether the documentation status continuum results in health disparities among immigrants by legal status, but the research does not find evidence of consistent disparities (for a review, see Hamilton, Hale, and Savinar 2019). This could be due to data constraints-particularly in the ability to sample undocumented immigrant populations and measure immigrant legal status in secondary data sources. Yet, studies using a range of methods and data sources arrive at similar conclusions (Bacong and Sohn 2020;Flores et al. 2012;Hamilton et al. 2019;Iten et al. 2014;Kelaher and Jessop 2002;Marshall et al. 2005;Ro and Hook 2021;Swartz et al. 2017;Wen and Maloney 2014). Existing studies analyze a wide variety of data sources, including surveys, insurance claims, driver's license applications, medical records, birth records, and community samples.

Because immigrant legal status is not commonly measured in population health data sources owing to its sensitive nature, many studies impute or use proxies for immigrant legal status, but other studies directly measure immigrant legal status in small, non-random samples of immigrants or in large, random, but unique samples of immigrants, such as farm workers.

Studies also analyze a diverse set of measures of health, including chronic illness prevalence, incidence of disease and symptoms, clinical measures of high blood pressure and diabetes outcomes, self-reported health, and birth outcomes. Regardless of differences in data and measurement, studies find few consistent legal status disparities in health.

As described above, the rights granted by one's position within the documentation status continuum can affect health through multiple mechanisms, especially socioeconomic status, health care, and the stress of immigration enforcement. Life course and immigrant integration theories suggest that while barriers to socioeconomic resources and health care are similar for 1 stand 1.5-generation immigrant adults, stress may function differently across generations, given differences in the primary institutions governing their lives (Abrego 2006;Enriquez 2020;Gleeson and Gonzales 2012;Gonzales 2011), resulting in potentially different patterns of health between and within generations. For example, all children are protected by equal access to public K-12 education and, in some states like California after 2016, to state-funded health insurance coverage, regardless of their immigration status (Gleeson and Gonzales 2012). However, the institutions governing adult life-most notably, the labor market, but also health care-use immigrant legal status as a barrier to adult entry, regardless of an immigrant's level or location of education attainment. Although undocumented 1.5-generation immigrants are educated in U.S. schools that foster a norm of meritocracy, legal barriers to work mean that their ambitions, efforts, and talents in school are not rewarded in the labor market (Abrego 2011;Gleeson and Gonzales 2012;Gonzales 2011).

While material disadvantages-restricted socioeconomic opportunities and access to health care-likely undermine health in similar ways for the 1 st and 1.5 generations, the stress of immigration enforcement may be differently harmful for the 1.5 generation, who immigrated as children. Immigrants who arrive as children grow up and form primary attachments and identities in the United States (Abrego 2011;Gleeson and Gonzales 2012;Patler 2018b;Vaquera et al. 2017). Unlike their parents, 1.5-generation immigrants do not commonly have a frame of reference based in the origin country with which to compare and evaluate their experiences as immigrants. Instead, peers, institutions, and culture in the destination country form the frame of reference. 1.5-generation immigrants come to know local meanings regarding immigrationthose based in a discriminatory ideology, policies, and actions-as they grow up amidst and are subject to them.

Research has shown that the circumstances of growing up in the United States, and forming primary attachments and identities there, amplify the indignity and harms of undocumented status. The implications of undocumented status become manifest at the transition to adulthood, as adolescents exit the K-12 school system and begin to engage with the institutions that govern adult life in the United States (Gonzales 2011). Research on the experiences of undocumented 1.5-generation immigrants has found that "learning to be illegal" is akin to "awakening to a nightmare": one's identity, plans, hopes, and dreams confront the reality of formal exclusion from U.S. institutions and the daily threat of legal action (Gonzales 2011;Gonzales and Chavez 2012). Given the vast expansion of immigration law enforcement over the past several decades, recent research has also suggested that LPRs may also experience heightened legal status-related distress that could make their experiences more similar to those of undocumented immigrants than to naturalized citizens (Asad 2020;Patler 2018a).

At the same time that the transition to adulthood involves a shift into adult-governing institutions that formally exclude undocumented immigrants, it is also a critical period of development during which lifelong health habits and patterns are established. Torres and Young (2016) applied life course theory to understand how the consequences of immigrant legal status on health may be especially harmful during the transition to adulthood. Adolescence and early adulthood are periods during which individuals establish health habits, such as smoking, and in which health status is often set, with lifelong consequences for health and socioeconomic status (Harris 2010). The coupling of the transition to adulthood with "awakening" to legal exclusion may therefore be especially harmful for the health of 1.5-generation non-citizen immigrants.

We ask two distinct research questions in this study. First, we ask whether there are legal status disparities within the 1 st and the 1.5 generations of adult Latinx immigrants in California.

Because legal status determines one's rights, including access to socioeconomic opportunities and health care, and stress imposed by the threat of immigration enforcement, we expect to observe legal status disparities in health among both generations. Second, we ask whether there are greater legal status disparities in health among the 1.5 than among the 1 st generation of adult Latinx immigrants in California. Because the 1.5 generation grows up in the same society that denies them membership and equal rights, we expect the harms of non-citizen immigration status to be greater for the health of 1.5 generation immigrants than the harms of non-citizen immigration status among the 1 st generation.

We use data from the 2005-2017 CHIS, a telephone survey of California's residential, non-institutionalized population collected by UCLA's Center for Health Policy Research (CHPR). The CHIS uses a dual-frame, multi-stage sampling design that draws from landline and cell phone numbers in California. The survey randomly selects one adult (age 18 and older) from each sampled household. Our analytical sample includes 23,288 Latinx immigrant adult (ages 18-65) respondents interviewed between 2005-2017, the years for which our key variables were consistently asked in the CHIS. Latinx immigrants are respondents who report that they were born outside the United States in countries in Latin America or the Spanish-speaking Caribbean.

We define the 1.5 generation as immigrants who arrived to the United States at age 12 or younger because primary socialization occurs prior to adolescence (Rumbaut 2004). The majority of Latinx immigrants in the CHIS are 1 st generation (82%, n=19,123), but by combining 12 years of data, we obtain a sufficient sample of 1.5-generation immigrants (n=4,165). We use survey weights to adjust for the complex sample design. The CHPR Data Access Center imputed missing values on all variables used in our analysis.

The CHIS is especially useful for our purposes because it includes a more detailed series of questions to determine immigrant legal status than national health surveys such as the National Health Interview Survey or the Behavioral Risk Factor Surveillance System. The CHIS asks foreign-born respondents first if they are citizens; then it asks those who are not citizens if they are LPRs. 3 Respondents who answer no to both questions include undocumented immigrants, refugees who have not yet adjusted status to LPR, temporary visa holders, and immigrants with discretionary status (such as DACA). The large majority of this group is likely undocumented; in prior years, studies estimated that 88-93% of Latinx non-citizen-non-LPRs in California were undocumented (Vargas Bustamante et al. 2012). 4 Many studies have used the CHIS to examine the health status and access to health care of undocumented immigrants (e.g. Bacong and Sohn 2020;Oropesa et al. 2015;Patler et al. 2019;Pourat et al. 2014;Vargas Bustamante et al. 2012).

California in this period provides a useful context for studying immigrant health disparities because it is the state with the largest immigrant population (twice as large as the second-largest state immigrant population, in Texas) and the greatest share of the population that is of immigrant origin (27%, followed by 23% in New Jersey and 22% in New York) (Migration Policy Institute 2021). In California, 48.5% of immigrants are of Latinx origins and 46.4% are non-citizens (MPI 2021). Population data from California will therefore provide a large sample of Latinx immigrants with substantial variation across legal statuses-as is true in the CHIS.

Still, as a context within which to study the harms of immigration enforcement, the state in this period is unique for its efforts to extend services and protections to undocumented immigrants.

Since 2000, the CA Legislature passed a number of laws expanding immigrant rights in this period, including, for example, the 2011 California Dream Act, which gave access to state funds for college tuition assistance for undocumented residents; 2013's AB60, which allows undocumented immigrants to apply for driver's licenses; and the 2014 California TRUST Act, which limited the collaboration between local law enforcement and federal immigration enforcement. In this context, it may be that non-citizen immigration status is not as harmful to health as in other states, such as Arizona or Alabama, where state law has taken the opposite approach. In other words, our estimates of group differences may not generalize beyond California or may be conservative estimates of group differences in less welcoming state climates.

We analyze a broad set of measures of health. Population health science does not give preference to particular health outcomes for the analysis of social inequality or social determinants of health (Hummer and Hamilton 2019). Furthermore, central theories of health disparities, such as fundamental cause theory, emphasize that social disparities in health vary consistently across a diverse set of outcomes (Link and Phelan 1995). It is also important to measure a diverse set of outcomes because of the unique biases that different measures of health present. For instance, in survey data such as the CHIS, questions about chronic conditions ask if the condition has been diagnosed by a health professional, which may result in undermeasurement among groups with less access to health care, such as undocumented immigrants (Pourat et al. 2014). Other measures, such as self-rated health, may reflect language and cultural differences in health assessment (Finch and Vega 2003). With this guidance in mind, we make use of all questions regarding physical and mental health available in the CHIS in the 2005-2017 waves. We measure four chronic conditions by respondents' reports of whether a health care professional has ever told them they have: (1) heart disease, (2) high blood pressure, (3) asthma, and (4) diabetes. We measure three current conditions: (5) asthma (whether respondent reports having had an episode of asthma or an asthma attack in the past 12 months), ( 6) one or more visits to the emergency room in the past year, and ( 7) obesity (measured as a body mass index (BMI), or reported weight in kilos divided by reported height in meters squared, of 30 or greater). These measures are useful supplements to chronic conditions because they capture current health and do not depend on a doctor diagnosis. We also analyze (8) self-rated health as a global measure of physical and mental health. Respondents reported whether their health was excellent, very good, good, fair, or poor. We dichotomized the 5-point scale into a binary variable (1= fair or poor health; 0= good, very good, or excellent health).foot_2 Finally, we use (9) the respondent's score on the Kessler 6-question (K6) psychological distress scale to analyze a dichotomous measure of moderate or severe distress (K6=>5) (Prochaska et al. 2012). The K6 scale includes six questions with Likert-scale responses assessing the recent frequency of mental health symptoms such as nervousness, hopelessness, and restlessness. The score is a count that ranges from zero to 24 points. 6We analyze each of the nine binary outcomes using linear probability models.

Coefficients from linear probability models can be interpreted as the effect of a one-unit change in independent variable on the probability of the outcome. 7 We use coefficients from the first model, which controls for age and survey year, to predict the probability of each outcome for each generation-legal status group, which we present in a series of graphs. Age controls account for the younger age distribution of undocumented immigrants, as compared to LPRs and naturalized citizens (see Table 1).

We then present results from a model that additionally controls for demographic characteristics, socioeconomic status, and access to health care. With this model, we assess whether differences across legal status and generational groups in these measured characteristics-especially socioeconomic status and health care-account for group differences in health. The model adds controls for characteristics that research has shown are associated with health, generation, and immigrant status, including gender, marital status (married versus not), education (less than high school, high school or GED, some college, or BA or higher), the household income to poverty ratio (expressed as a percentage of the federal poverty line), Mexican ethnicity, current insurance coverage (public or private, versus none), and having a usual source of health care (Bacong and Sohn 2020;Finch and Vega 2003;Pourat et al. 2014;Ro and Hook 2021;Vargas Bustamante et al. 2012;Young and Pebley 2017). We present these results as regression coefficients and as average marginal effects for key group contrasts.

Table 1 shows the average characteristics of each of the six groups of Latinx immigrant adults in the CHIS between 2005-2017 (naturalized citizens, LPRs, and undocumented, across the 1 st and 1.5 generations). The table reveals that there are significant differences in age, marital status, education, income, years in the United States, Mexican ethnicity, Spanish language, insurance coverage, and usual source of care between and within immigrant generations by legal status. On a whole, undocumented immigrants and LPRs are less advantaged than naturalized citizens, and 1 st -generation immigrants are less advantaged than 1.5-generation immigrants. For example, two-thirds of undocumented 1 st -generation Latinx immigrants have less than a high school education, a significantly higher share than 1 st generation citizens (half of whom have a less than a high school education). However, all groups of 1.5-generation Latinx immigrants are less likely to have less than a high school education than 1 st -generation citizens: about a quarter of undocumented and LPR 1.5-generation immigrants have less than a high school education, and 12.6% of 1.5-generation citizens have less than a high school education. A second finding of note is that LPRs and undocumented immigrants tend to be more similar to each other than either group is to citizens. Within both generations, citizens are older, are more likely to be married, are better educated, have higher income, have been in the United States more years, are less likely to be Mexican, are less likely to answer the interview in Spanish, are more likely to be insured, and are more likely to have a usual source of health care, than the other two groups.

Figure 1 shows age-and year-adjusted predicted rates (probabilities estimated from Model 1 in Table 3, converted to percentages) of nine health outcomes among adult Latinx immigrants in California across legal status and generation. Consistent with previous research, patterns of health outcomes do not vary in an obvious or consistent way across outcomes.

However, some tentative conclusions can be drawn from the graphs. There are higher predicted rates of some outcomes-heart disease, high blood pressure, diabetes, poor health, and mental distress-among non-citizens (LPRs and undocumented immigrants) than citizens, though in some instances the confidence intervals are overlapping. For heart disease, citizens in both generations have lower rates than non-citizens; in the 1 st generation, 2.8% of citizens report ever being diagnosed with heart disease, compared to 4.4% of LPRs and undocumented immigrants.

In the 1.5 generation, the predicted rates are slightly higher, with the highest rate among LPRs (at 5.5%), followed by undocumented immigrants (4.6%) and citizens (3.8%). Poor self-rated health shows a legal status gradient within both generations: 37% of undocumented 1.5generation immigrants and 42% of undocumented 1 st -generation immigrants rate their health as fair or poor (compared to good, very good, or excellent), compared to 33% of 1.5-generation and 38% of 1 st -generation LPRs and 21% of 1.5-generation and 27% of 1 st -generation citizens. We also observe a legal status gradient in mental distress in the 1 st generation, but in the 1.5 generation, the larger difference is between citizens and non-citizens.

Table 2 provides statistical tests of differences in estimates (average marginal effects) across key groups, and Table 3 shows regression coefficients for the base and interacted terms (coefficients for covariates not shown). Table 2 shows statistically significant legal status differences in health on some outcomes within each generation, providing some support for our first research expectation of within-generation legal status disparities in health. Within the 1 st generation, LPRs and undocumented immigrants have significantly higher rates of heart disease, poor health, and distress than citizens. Undocumented immigrants also have higher rates of poor health than LPRs, but otherwise these two groups of non-citizens have similar rates of health within the 1 st generation. Within the 1.5 generation, there are differences in poor health and distress, as well as in high blood pressure and diabetes, between citizens and non-citizens. In the 1.5 generation, LPRs and undocumented immigrants have similar rates of all nine measures of health.

Our second expectation was that legal status differences in health would be greater among the 1.5 generation than among the 1 st . The interaction term between immigrant generation and legal status in the models shown in Table 3 provides a test of this expectation. None of the interaction terms are significant for any outcome. In other words, the differences we observe in health across legal statuses in the 1 st generation are not significantly different from the differences in health across legal status in the 1.5 generation.

Model 2 in Table 3 controls for all measures in Table 1, with key contrasts shown in Panel B of Table 2. The two sets of results show that nearly all group differences are reduced in size and/or are imprecisely estimated in this model. This finding suggests that most group differences in health observed in Figure 1 and Model 1 are driven by group differences in the characteristics included in the models and their associations with health.

In this study, we examined the association between immigrant legal status and nine health outcomes among 23,288 Latinx immigrant adults in the 2005-2017 CHIS, considering differences within and between the 1 st and 1.5 immigrant generations. Non-citizen immigrants in the United States face substantial socioeconomic and health care disadvantages that arise from the denial of rights and the significant threat of detention and deportation, and these forces arguably undermine immigrant health. We expected to find legal status differences in health within both generations, but we also expected that the health consequences of noncitizen status would be greater among 1.5-generation immigrants, who grew up in the same society that denies them formal membership.

We find some evidence of legal status disparities in health within the 1 st and 1.5 generations of Latinx immigrant adults in the United States. In the 1 st generation, non-citizens have higher rates of heart disease, while in the 1.5 generation, non-citizens have higher rates of high blood pressure and diabetes. In both generations, non-citizens have higher rates of mental distress and poor self-rated health than citizens. We did not find support for our second expectation, which was that legal status disparities in health would be greater among the 1.5 generation. While we do not observe significant generational differences in the association between legal status and health, we do observe different patterns across physical health outcomes. Because the generational patterns of legal status disparities in physical health vary across outcomes, analyses that combine the 1 st and 1.5 generations may obscure legal status differences in physical health.

Within both generations, we observe differences in health between citizens and noncitizens, but not among non-citizens. Other studies have found similar patterns in educational and labor market settings (Patler 2018a;Patler, Gleeson, and Schonlau 2020). Although LPR status provides immigrants with greater rights and protections than available to undocumented immigrants, integration into U.S. institutions, including health care, may place immigrants with more stable legal status at risk. Asad (2018) finds that LPRs in Dallas perceive greater vulnerability to deportation than undocumented immigrants due precisely to their system embeddedness, or formal involvement with U.S. institutions. In other words, the inclusion that theoretically gives immigrants with greater legal rights more stability and support in the United States at the same time make LPRs feel more vulnerable. The U.S. immigration regime may be as detrimental to the sense of security and wellbeing of immigrants with permanent resident visas, even though permanent resident visas provide fundamental material and legal protections.

Although we find some evidence of legal status disparities in health on some measures, we do not observe disparities on others. This could reflect the unique context of California in the 2005-2017 period, during which the state passed a number of laws expanding immigrant rights which may have mitigated some disparities we might have otherwise have observed. It is also possible that the absence of consistent health disparities across legal status reflects the racialization of legal status. As Asad and Clair (2018) have argued, undocumented immigration status is racialized to disproportionately impact Latinx immigrants, such that immigration law and enforcement, and the discourse surrounding immigration law, have both direct effects on the health of undocumented immigrants and spillover effects onto documented Latinx immigrants, as well as onto U.S.-born Latinx communities. If this is the case, then legal status as a racialized fundamental cause of health may suppress within-generation differences by legal status. In other words, we may not see a consistent undocumented or non-citizen disadvantage in physical health because the consequences of the discriminatory and enforcement regime against undocumented and other non-citizen immigrants in the United States affect all members of the Latinx immigrant community.

We also find that socioeconomic status and health care fully mediate within-generation legal status disparities in physical health and mental distress. This finding suggests that legal status stratification in socioeconomic opportunities and health care are primary mechanisms through which immigration policy and enforcement undermine immigrant health (see also Bacong and Sohn 2020). However, these findings do not rule out the substantial role of stress generated by immigration law and enforcement. For one, disparities in poor self-rated health among the 1 st generation persist in spite of controls for socioeconomic and health care, which could indicate the independent role of stress. At the same time, stress arguably impacts the covariates in the model, as multiple studies show that fear of immigration enforcement and stress generated by legal uncertainty and exclusion affect interactions in family life, with health care institutions and providers, in school, and at work (Cardoso et al. 2021;Enriquez et al. 2018;Gándara and Ee 2021;Gomez Cervantes and Menjívar 2020;Valdivia 2019). In other words, group differences in these mediators may themselves be outcomes of the stress and fear generated by immigration enforcement, and not independent mechanisms of health disparities.

We find greater evidence of a legal status gradient in health than other studies. This could be because prior studies do not disaggregate by immigrant generation. However, the difference may also owe to study design, including the population studied and the data and key measures used. Two studies using national data-one on Mexican-born farm workers and one on Latinx adults-found evidence of the opposite pattern to ours, that is, of better health among less legally privileged immigrants (Hamilton et al. 2019;Ro and Van Hook 2021). It is possible that the difference owes to differences in the composition of immigrants across U.S. states. There may be unique patterns of selectivity into different state destinations, reflecting distinct costs and networks available to migrants. Testing state variation in the association between legal status and health is one of many steps for future research on this topic. Another important step for research is to incorporate dynamic, retrospective measures of legal status. This study and most others documenting inequality in immigrant health rely on static measures of legal status, but immigrant legal status frequently changes (Flores Morales 2021). Research on DACA shows that the expansion of legal rights had short-term health benefits for participants and their children (Patler et al. 2019;Patler, Hamilton, and Savinar 2019). Future research on legal status disparities in immigrant health should also account for changes to legal status across the life course, and how those may influence health.

Table 1. Weighted, age and survey year adjusted means and percent distributions of demographic and social characteristics of Latinx adults in California, 2005-2017, by legal status, within the 1 st and 1.5 generations Naturalized citizen LPR Undocumented Panel A. 1 st generation Age (mean) 49.7* a 44.4* a 37.8* a Male (%) 46.7* 50.4 51.8* Married (%) 73.0* a 65.5* a 44.4* a Education (%) Less than HS 50.4* a 65.4 a 66.7* a HS/GED 23.9* a 18.9* a 21.6 a Some college 13.0* a 8.0 a 6.5* a BA or higher 12.7* a 7.7* 5.2* Income-to-poverty ratio (mean) 2.5* a 1.7* a 1.2* a Years in the U.S (mean) 27.3* a 20.4* a 13.5* a Mexican (%) 73.0* 79.5 80.2* a Interview in Spanish 66.0* a 85.9* a 92.4* a Currently insured (%) 79.4* a 69.7* a 48.3* Has usual source of care (%) 81.1* 70.9* 58.5* Sample size 6,314 6,447 6,362 Panel B. 1.5 generation Age (mean) 38.4* 33.0* 26.1* Male (%) 49.0 53.1 44.3 Married (%) 51.4* 35.3* 24.2* Education (%) Less than HS 12.6* 27.3 27.5* HS/GED 29.4 29.7* 48.9* Some college 31.7 31.8* 19.7* BA or higher 26.3* 11.2* 3.9* Income-to-poverty ratio (mean) 3.6* 2.4* 1.6* Years in the U.S (mean) 32.8* 26.5* 19.6* Mexican (%) 73.6* 82.5* 91.3* Interview in Spanish 9.3* 26.1 31.0* Currently insured (%) 85.0* 76.0* 52.2* Has usual source of care (%) 82.8* 70.8 64.3* Sample size 2,723 998 444 *indicates mean is statistically different at p<0.05, comparing naturalized citizens to LPRs; LPRs to undocumented; and undocumented immigrants to naturalized citizens within each generation a indicates mean is statistically different at p<0.05, comparing 1 st to 1.5 generation within each legal status group Table 2. Marginal effects on nine health outcomes by generation within legal status and by legal status within generation, among Latinx immigrant adults in California, 2005-2017 Heart Disease High Blood Pressure Diabetes Lifetime Asthma Current Asthma ER Visit Obesity Poor Health Distress AME SE AME SE AME SE AME SE AME SE AME SE AME SE AME SE AME SE Panel A (Model 1) Within legal status Citizen: 1st vs 1.5 1.02 0.79 -1.16 1.65 -1.15 1.17 2.75* 1.14 1.22 0.78 4.20* 1.70 0.24 2.01 -5.60** 1.77 -2.91 1.95 LPR: 1st vs 1.5 1.05 1.10 3.11 2.42 1.46 1.78 2.09 1.42 1.19 1.04 4.81* 2.14 3.32 2.73 -5.05* 2.56 -0.18 2.65 Undoc: 1st vs 1.5 0.17 0.89 -2.35 1.90 0.03 1.16 3.71 2.63 2.29 2.07 0.74 2.87 -0.92 3.30 -5.02 3.11 -3.91 3.59 Within generation 1st: Citizen vs LPR 1.68* 0.68 1.18 1.44 2.06 1.16 -0.71 0.76 -0.32 0.56 -1.37 1.14 0.61 1.58 11.13** 1.56 3.54* 1.54 1st: Citizen vs Undoc 1.67* 0.69 2.58 1.46 1.39 1.09 -1.45 0.81 -0.44 0.62 -0.65 1.24 2.17 1.68 15.06** 1.61 6.20** 1.61 1st: LPR vs Undoc -0.01 0.60 1.41 1.38 -0.66 1.06 -0.74 0.75 -0.12 0.55 0.73 1.12 1.56 1.57 3.93* 1.58 2.66 1.56 1.5: Citizen vs LPR 1.71 1.10 5.44* 2.47 4.66* 1.73 -1.37 1.62 -0.35 1.14 -0.77 2.40 3.69 2.90 11.68** 2.61 6.28* 2.81 1.5: Citizen vs Undoc 0.82 0.95 1.40 2.02 2.57* 1.23 -0.50 2.72 0.63 2.10 -4.11 3.05 1.02 3.46 15.63** 3.16 5.20 3.73 1.5: LPR vs. Undoc -0.89 1.20 -4.05 2.71 -2.09 1.74 0.87 2.84 0.98 2.20 -3.34 3.31 -2.67 3.87 3.95 3.60 -1.08 4.04 Panel B (Model 2) Within legal status Citizen: 1st vs 1.5 1.49 0.93 -2.06 2.07 -0.57 1.50 0.45 1.38 -0.18 0.94 3.31 1.95 2.11 2.40 4.22 2.19 -0.61 2.32 LPR: 1st vs 1.5 1.39 1.14 1.53 2.72 1.33 2.04 -0.34 1.59 -0.34 1.20 3.16 2.33 4.16 3.07 3.73 2.99 1.27 2.99 Undoc: 1st vs 1.5 0.65 1.08 -3.92 2.37 0.01 1.55 1.53 2.59 0.81 2.02 -0.12 2.91 -1.64 3.54 4.04 3.32 -1.96 3.86 Within generation 1st: Citizen vs LPR 1.27 0.70 1.11 1.51 1.22 1.20 0.31 0.77 0.31 0.58 -0.55 1.18 -0.91 1.66 4.75** 1.60 1.00 1.62 1st: Citizen vs Undoc 0.99 0.79 3.21 1.62 0.65 1.18 0.31 0.93 0.64 0.75 1.13 1.38 0.81 1.86 4.86* 1.78 1.17 1.80 1st: LPR vs Undoc -0.28 0.64 2.10 1.39 -0.57 1.04 0.00 0.80 0.33 0.62 1.67 1.15 1.72 1.59 0.10 1.55 0.17 1.56 1.5: Citizen vs LPR 1.16 1.12 4.70 2.48 3.12 1.71 -0.48 1.65 0.15 1.15 -0.69 2.42 1.14 2.88 4.27 2.70 2.89 2.83 1.5: Citizen vs Undoc 0.15 0.96 1.36 2.17 1.23 1.36 1.39 2.72 1.63 2.13 -2.30 3.03 -2.94 3.51 4.68 3.11 -0.18 3.82 1.5: LPR vs. Undoc -1.02 1.21 -3.34 2.71 -1.89 1.75 1.87 2.78 1.48 2.19 -1.61 3.23 -4.08 3.82 0.41 3.55 -3.07 4.08 Bolded and ** p<0.01, * p<0.05

Table 3. Linear probability regression coefficients of nine health outcomes by immigrant legal status, generation, and their interaction among Latinx immigrant adults in California, 2005-2017 Model 1 Age + Year Model 2 All covariates Coeff. SE Coeff. SE 1. Heart disease 1.5 generation (ref = 1 st ) 0.010 (0.008) 0.015 (0.009) LPR (ref = citizen) 0.017* (0.007) 0.013 (0.007) Undoc 0.017* (0.007) 0.010 (0.008) LPR * 1.5 0.000 (0.013) -0.001 (0.013) Undoc*1.5 -0.009 (0.012) -0.008 (0.012) 2. High blood pressure 1.5 generation (ref = 1 st ) -0.012 (0.017) -0.021 (0.021) LPR (ref = citizen) 0.012 (0.014) 0.011 (0.015) Undoc 0.026 (0.015) 0.032* (0.016) LPR * 1.5 0.043 (0.029) 0.036 (0.028) Undoc*1.5 -0.012 (0.024) -0.019 (0.025) 3. Diabetes 1.5 generation (ref = 1 st ) -0.012 (0.012) -0.006 (0.015) LPR (ref = citizen) 0.021 (0.012) 0.012 (0.012) Undoc 0.014 (0.011) 0.007 (0.012) LPR * 1.5 0.026 (0.021) 0.019 (0.020) Undoc*1.5 0.012 (0.016) 0.006 (0.016) 4. Ever asthma 1.5 generation (ref = 1 st ) 0.028* (0.011) 0.005 (0.014) LPR (ref = citizen) -0.007 (0.008) 0.003 (0.008) Undoc -0.015 (0.008) 0.003 (0.009) LPR * 1.5 -0.007 (0.018) -0.008 (0.018) Undoc*1.5 0.010 (0.028) 0.011 (0.027) 5. Current asthma 1.5 generation (ref = 1 st ) 0.012 (0.008) -0.002 (0.009) LPR (ref = citizen) -0.003 (0.006) 0.003 (0.006) Undoc -0.004 (0.006) 0.006 (0.008) LPR * 1.5 -0.000 (0.013) -0.002 (0.013) Undoc*1.5 0.011 (0.022) 0.010 (0.021)

Table 3 is continued on next page

Table 3 continued Model 1 Age + Year Model 2 All covariates Coeff. SE Coeff. SE 6. ER visit 1.5 generation (ref = 1 st ) 0.042* (0.017) 0.033 (0.020) LPR (ref = citizen) -0.014 (0.011) -0.006 (0.012) Undoc -0.006 (0.012) 0.011 (0.014) LPR * 1.5 0.006 (0.027) -0.001 (0.026) Undoc*1.5 -0.035 (0.032) -0.034 (0.031) 7. Obesity 1.5 generation (ref = 1 st ) 0.002 (0.020) 0.021 (0.024) LPR (ref = citizen) 0.006 (0.016) -0.009 (0.017) Undoc 0.022 (0.017) 0.008 (0.019) LPR * 1.5 0.031 (0.033) 0.021 (0.032) Undoc*1.5 -0.012 (0.038) -0.038 (0.037) 8. Poor health 1.5 generation (ref = 1 st -0.056** (0.018) 0.042 (0.022) LPR (ref = citizen) 0.111*** (0.016) 0.048** (0.016) Undoc 0.151*** (0.016) 0.049** (0.018) LPR * 1.5 0.006 (0.030) -0.005 (0.030) Undoc*1.5 0.006 (0.035) -0.002 (0.033) 9. Distress 1.5 generation (ref = 1 st ) -0.029 (0.020) -0.006 (0.023) LPR (ref = citizen) 0.035* (0.015) 0.010 (0.016) Undoc 0.062*** (0.016) 0.012 (0.018) LPR * 1.5 0.027 (0.032) 0.019 (0.032) Undoc*1.5 -0.010 (0.040) -0.014 (0.040) *** p<0.001, ** p<0.01, * p<0.05 2.8 4.4 4.4 3.8 5.5 4.6 0 1 2 3 4 5 6 7 8 Cit LPR Undoc Cit LPR Undoc 1st generation 1.5 generation Heart Disease 23.6 24.8 26.2 22.4 27.9 23.8 0 5 10 15 20 25 30 35 Cit LPR Undoc Cit LPR Undoc 1st generation 1.5 generation High Blood Pressure 11.6 13.7 13.0 10.5 15.1 13.0 0 4 8 12 16 20 Cit LPR Undoc Cit LPR Undoc 1st generation 1.5 generation Diabetes 7.2 6.4 5.7 9.9 8.5 9.4 0 2 4 6 8 10 12 14 16 Cit LPR Undoc Cit LPR Undoc 1st generation 1.5 generation Lifetime Asthma 3.7 3.4 3.3 4.9 4.6 5.6 0 2 4 6 8 10 Cit LPR Undoc Cit LPR Undoc 1st generation 1.5 generation Current Asthma 33.1 33.8 35.3 33.4 37.1 34.4 0 5 10 15 20 25 30 35 40 45 Cit LPR Undoc Cit LPR Undoc 1st generation 1.5 generation Obesity 16.2 14.9 15.6 20.4 19.7 16.3 0 5 10 15 20 25 Cit LPR Undoc Cit LPR Undoc 1st generation 1.5 generation ER Visit 26.8 37.9 41.9 21.2 32.9 36.9 0 5 10 15 20 25 30 35 40 45 50 Cit LPR Undoc Cit LPR Undoc 1st generation 1.5 generation Poor Health 31.7 35.2 37.9 28.8 35.0 34.0 0 5 10 15 20 25 30 35 40 45 Cit LPR Undoc Cit LPR Undoc 1st generation 1.5 generation Mental Distress