Immigration concerns and social program avoidance: The roles of legal status and family composition among asian and latino communities

Highlights

• •We study legal status, family composition, and ethnicity, and public program avoidance.
• •Legally precarious immigrants were more likely to avoid public programs.
• •Asian lawful permanent residents were likelier to avoid public programs than Latinos.

Abstract

Objectives

Immigration policies significantly affect immigrants and their families' access to social programs. This study examines the role of legal status and familial composition in Asian and Latino adults’ avoidance of social programs and assesses differences between the groups. We categorized respondents’ familial composition based on whether all household members had the same citizenship status. We created two groups: respondents with cross-status ties and those without.

Methods

We use data from 1000 U.S. adults with proximal or distant ties to noncitizens (collected in September 2021) to employ multivariable generalized linear models with binomial family and logit link to assess avoidance of needed social programs due to immigration concerns.

Results

Our fully adjusted model reveals that compared to U.S. citizens without familial cross-status ties (i.e., all-citizen household members), legally precarious immigrants (LPI) without cross-status ties (odds ratio (OR)= 3.64, 95 % CI: 1.67–7.96), LPI with cross-status ties (OR=1.71, 95 % CI: 1.14–2.57), and U.S. citizens with cross-status ties (OR=1.66, 95 % CI: 1.14–2.40), were more likely to report avoidance of needed social support programs. Further, an interaction analysis shows that Asian lawful permanent residents with cross-status ties exhibit a higher likelihood of avoiding social programs due to immigration concerns compared to their Latino counterparts.

Conclusion

The far-reaching consequences of anti-immigrant policies on noncitizen individuals and their families, including U.S. citizens, may lead to disparities in access to social assistance programs that may exacerbate health disparities. Further, our findings suggest that immigrants’ cross-status ties to U.S. citizen family members may be protective against the harmful effects of anti-immigrant policies.

Introduction

Social safety-net programs for food, housing, and healthcare have been shown to improve health and social outcomes (Bailey et al., 2024; Gassman-Pines and Hill, 2013). For example, food assistance programs like the Supplemental Nutritional Assistance Program (SNAP) in the United States (U.S.) are associated with reduced likelihood of food insecurity, better self-rated health, and better cognitive performance (Gundersen and Ziliak, 2015; Bitler and Seifoddini, 2019). Similarly, participation in housing assistance programs, including vouchers and public housing, also improves health and psychological well-being among U.S. adults (Fenelon et al., 2017). Further, studies have shown that Medicaid coverage led to increased preventive care, better chronic disease management, and improved financial well-being (Allen and Sommers, 2019). However, immigrant communities in the U.S. face persistent barriers to accessing these health-enhancing social programs due to exclusionary policies (Gee and Ford, 2011; Menjívar, 2023).

One of these policies is public charge inadmissibility rule. Public charge was initially enacted through the Immigration Act of 1882 to deny entry and the ability to adjust their status to immigrants who cannot support themselves (Pillai and Published, 2022). Undocumented immigrants deemed a public charge to the U.S. government cannot adjust their legal status to lawful permanent residents. Historically, cash assistance and long-term care were the only programs considered in the public charge determination. In September 2018, however, the Trump Administration announced its plan to broaden the definition of who could be labeled a public charge by including housing, food, and healthcare programs. The public charge expansion took effect in February 2020, a month before the COVID-19 pandemic became a national emergency. Following Trump's public charge expansion, noncitizens used healthcare and food social programs significantly less (Barofsky et al., 2020). Despite the Biden Administration's reversal of the public charge rule in March 2021, fear and misinformation among the immigrant community persist, creating a “chilling effect” in which immigrants and their families avoid social assistance programs over the fear of immigration-related consequences (Wolwowicz-Lopez et al., 2023).

The chilling effect refers to immigrants’ behavior changes and avoidance of benefits they and their families are eligible for due to fears of jeopardizing their future adjustment of status (Young et al., 2023). For instance, the chilling effect hypothesis has been used to explain ecological-level changes in safety-net enrollment across various groups before and after the rule change announcement. Using county-level data, studies find that the announcement of the public charge expansion reduced children's access to safety-net benefits, including Medicaid, SNAP, and Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) (Barofsky et al., 2020). Other studies characterize which immigrants are likely to forgo enrolling in social programs due to sanctions following the rule, focusing on essential workers and their families (Touw et al., 2021). More recently, a California-based study comparing immigrants with and without lawful permanent residency (LPR) to U.S. citizens applied estimates of public benefit avoidance from an Urban Institute survey to estimate the number of Latino and Asian immigrants that would potentially forgo Medicaid (Bustamante et al., 2022); they estimated that up to 192,905 Latino and 4702 Asian immigrants in California without LPR would be at risk of losing their Medicaid insurance. Although innovative, these studies fail to inquire directly about immigrants’ experiences with social program avoidance due to immigration fears and instead rely on ecological trends in program utilization. Further, scholars have yet to consider crucial factors that shape how immigrants and their families respond to public policy changes like public charge, such as an individual's awareness of changes to the rule, understanding of social programs that could result in someone being classified as a public charge, and variations in family composition across legal status.

Family composition and cross-status relationships

Legal status, though granted to individuals, profoundly impacts families. The average undocumented individual has resided in the U.S. for over 20 years, and 70 % of them are part of mixed-status families—households with members holding varying citizenship and legal statuses (Enriquez, 2015). Consequently, the experiences of illegality—the condition of lacking lawful legal status—faced by one family member may have repercussions for the family unit (Rodriguez, 2016). Restrictive immigration policies affect not only undocumented individuals but also their citizen relatives, potentially limiting their access to health-enhancing social programs. Previous work has focused on how immigration laws impact the children of undocumented immigrants in the U.S. due to their reliance on and daily interactions with their parents (Dreby, 2015). Qualitative studies conducted before the 2018 public charge rule expansion have found that Latino immigrant families often avoided healthcare due to fears of immigration-related ramifications (Galletly et al., 2023), underscoring the broader effects of immigration policies on family well-being.

Nevertheless, it remains unclear whether legally precarious immigrants (LPI), such as undocumented immigrants and those with temporary statuses, benefit from their familial ties to U.S. citizens in navigating anti-immigrant policies. U.S. citizen family members can offer social support, information, and economic resources to their noncitizen relatives, potentially mitigating the adverse effects of illegality and social exclusion (Rodriguez, 2019). While cross-status family ties cannot entirely shield noncitizens from the legal precarity associated with liminal legal statuses, they may improve immigrants’ willingness to use social programs in exclusionary social environments toward immigrants through various forms of support. This includes informational support (e.g., advice, situation appraisal, and resource direction), tangible support (e.g., providing essential resources and assistance in applying to social programs), and extended networks (e.g., connections to community organizations or attorneys). Recent studies have found that undocumented young adults often rely on lawfully present family members to buffer the negative health impacts of economic insecurity, underscoring the significance of cross-status support from relatives with more secure legal statuses (Morales et al., 2024). The presence or absence of cross-status ties within immigrant families may influence their willingness to utilize social programs in anti-immigrant contexts.

Asian and latino immigrants’ experiences

Research on the impact of public charge expansion on immigrant families has predominantly focused on Latino communities, often neglecting the experiences of Asian immigrants (Wolwowicz-Lopez et al., 2023; Iraheta and Morey, 2023). It is essential to examine how ethnicity shapes the experience of illegality, particularly given the racialization of undocumented immigrants as Latinos in political and public discourse (Cho, 2017). As of 2017, 14–16 % of the 10.5 million undocumented individuals in the U.S. were from Asian countries (Budiman and Ruiz, 2023), meaning approximately one in six undocumented immigrants is of Asian origin. Between 2000 and 2013, the number of undocumented immigrants from Asia increased by 202 %, compared to a 29 % increase in those from Mexico (Capps et al., 2020). Additionally, the socioeconomic profiles of Asian and Latino immigrants differ significantly, with Asian immigrants generally having higher educational attainment, higher incomes, and, in some cases, better health outcomes than their Latino counterparts (Ro and Van Hook, 2021). Scholars suggest these socioeconomic and health differentials may be attributed to the groups’ mode of migration, such that Asian undocumented immigrants are more likely to have overstayed their tourist or student visas than their Latino counterparts (Ro and Van Hook, 2021). These socioeconomic and premigration differences may also influence the extent to which Asian immigrants rely on family members to alleviate financial strain, which may affect their engagement with social safety-net programs.

The social experiences of Latino and Asian immigrants in the U.S. also vary. Unlike Latino immigrants, Asian immigrants are often stereotyped as the “model minority,” which may protect them from being profiled as undocumented (Yi et al., 2016). Avoidance of social programs due to immigration concerns may be less prevalent among Asians, compared to Latinos, due to the groups’ lower likelihood of experiencing racialized stereotypes related to being undocumented. For instance, Asian immigrants may be more likely to engage with formal institutions and seek social assistance they may need as they experience less racialized stigma.

On the other hand, the legal invisibility of Asian immigrants may lead to intensified shame and isolation as a strategy to avoid revealing their undocumented status (Cho, 2017). While this legal invisibility may appear protective, it can also limit access to community networks and legal resources to combat the consequences of restrictive immigration policies. The limited resources and information may lead to Asian immigrants being less aware of the social programs they may qualify for or how to access services. While important, these competing hypotheses remain untested in the current literature. These findings highlight the need for a comprehensive and thoughtful approach to examine the diverse experiences of immigrants in the U.S. after the public charge expansion and reversal.

Research objectives and hypotheses

Given the heterogeneous experiences of immigrants during heightened anti-immigrant contexts, we seek to understand the extent to which individuals of distinct legal statuses, family compositions (i.e., presence or absence of a cross-status tie), and ethnic backgrounds experience a chilling effect following the 2018 public charge expansion. Using unique individual-level data of individuals with different legal statuses and U.S. citizens with proximal ties to noncitizens, we examine differences in avoidance of social programs due to immigration-related concerns across respondents’ legal status and proximal cross-status ties. We also assess whether the relationship between legal status, family composition, and foregone social assistance varies by self-reported ethnic background (Asian vs. Latino). We posit that regardless of cross-status ties, legally precarious immigrants (LPI), such as undocumented individuals, will be more likely to report avoidance of social assistance programs due to immigration concerns than U.S. citizens in all-citizen households (hypothesis 1a). We also hypothesize that LPI without familial cross-status ties to a U.S. citizen will be likelier to report they avoided social programs due to immigration concerns than an LPI respondent with cross-status ties (hypothesis 1b). Further, U.S. citizens with cross-status ties to a noncitizen family member will be more likely to report they avoided social programs due to immigration concerns than U.S. citizens without a familial cross-status tie (hypothesis 1c). Finally, we posit that differences in these associations for Asian or Latino respondents could go in either direction, given the racialized legal status of Latinos and Asian individuals’ legal invisibility (hypothesis 2).

Research methodology

The survey involved 1000 interviews with U.S. adults who are either connected to noncitizens or are noncitizen respondents. The instrument was designed by a panel of academic experts and leading immigrant and health rights advocates. Data was collected nationwide through online and live phone interviews from September 1 to 30, 2021. To maximize coverage, we worked with multiple online vendors with extensive and comprehensive databases of diverse Americans to garner a sample frame of U.S. adults who are likely to have proximal and distal connections to noncitizens; all respondents were randomly chosen to receive an invitation to participate in the study via telephone or online. Online invites were sent to potential participants via text-to-web, email invitation, and panel-listed sample self-administered online survey. At the same time, telephone interviews were conducted by live interviewer-assisted cell phone or landline phone; 28.5 % of respondents completed the survey over the phone, and the remaining 71.5 % online. To ensure the sample's representativeness, the survey was available in English, Spanish, Chinese (Mandarin and Cantonese), Korean, Vietnamese, and Tagalog. The data was weighted to ensure a demographic balance of the sample based on the demographic parameters defined by the 2020 American Community Survey's adult population with ties to noncitizens, including gender, education, age, and race/ethnicity. We excluded 40 respondents who selected “other” self-reported race/ethnicity, yielding an analytic sample of 960 Asian and Latino respondents. All data are at the individual level.

Measures

Dependent variable

To assess whether respondents avoided applying for social programs due to immigration concerns, we asked: “Over the past year and a half during the COVID-19 pandemic, a lot of people signed up for assistance programs, while other people felt uncertain if they were eligible to apply. Which of the following comes closest to you and your family?” Respondents selecting “myself or members of my family needed assistance but did not apply for assistance because of concerns over immigration status” were classified as 1, and 0 otherwise (i.e., if respondent or family applied for and received assistance or if respondent or family were never in need of any assistance programs). We combined the two groups who did not report avoidance of social programs due to immigration concerns, regardless of their self-reported need, because the study's primary interest is understanding the decision to avoid applying to social assistance programs due to immigration concerns.

Independent variable

A combination of respondents’ legal status and family composition was the independent variables of interest. This variable was composed of three items. First, respondents were asked about their nativity and whether they were born in the U.S., Puerto Rico, another U.S. territory, or another country. Respondents born in the U.S. or a U.S. territory were classified as U.S. citizens. Then, to measure legal status, respondents not born in the U.S. or U.S. territory were asked to identify whether they were currently any of the following: a) naturalized U.S. citizens, b) lawful permanent residents (LPR), c) Deferred Action for Childhood Arrivals (DACA), Temporary Protected Status (TPS), or work permit holders, or d) “something else.” Those who selected “something else” were considered undocumented immigrants or people without lawful status. Given the shared precarity and legal liminality of undocumented and DACA/work permit holders, we combined response options c) and d) into a single category of undocumented or legally precarious immigrants (Asad, 2020). We combined U.S. naturalized and native-born U.S. citizens as one category.

Next, we assessed family composition to differentiate between those with and without proximal familial cross-status ties. Respondents with proximal cross-status ties may be U.S. citizens with noncitizen family members, or vice versa, non-citizen respondents with U.S. citizen family members. Noncitizen family members may include undocumented individuals, lawful permanent residents, student- or work-visa holders, work permit/TPS holders, or DACA beneficiaries. On the other hand, respondents whose household members hold the same citizenship status as the respondents are classified as lacking proximal cross-status ties. We asked, “Which of the following best describes the people who currently live in your household?” Responses included: a) Everyone in my household is a U.S. citizen; b) some are U.S. citizens, and some are not; c) no one in my household is a U.S. citizen (option c was only offered to noncitizen respondents). We distinguish groups with no proximal cross-status ties (i.e., options a and c) from their counterparts (i.e., option b) to examine the influence of lacking such ties. We focus on ties between family members of different statuses because these proximal relationships are often the first line of support in the context of U.S. immigration policy changes, affecting access to resources, family experiences, and behaviors (Abrego, 2018).

Participants’ legal statuses and family composition were then classified into the following five mutually exclusive groups: a) U.S. citizen respondents without proximal cross-status ties, that is, all family members are U.S. citizens (reference group), b) U.S. citizen respondents with proximal cross-status ties to noncitizens, c) lawful permanent residents (LPR) with cross-status ties to citizens or noncitizens, d) legally precarious immigrants (LPI) with cross-status tie to U.S. citizens, e) and LPI without cross-status ties, that is, all family members are noncitizens.

We kept LPR as a separate noncitizen category since their experiences are distinct from those with liminal statuses, such as DACA and TPS beneficiaries, and from naturalized and U.S.-born individuals. Further, while some LPR (n = 12) individuals responded that they lived in a household where all members were U.S. citizens, they were categorized as having cross-status ties, where they were the noncitizen members.

Self-reported ethnicity was assessed with the following item: “Which of the following best describes your race or ethnicity?” Response options included Hispanic or Latino, White, not-Hispanic, Black or African American, Asian or Pacific Islander, Native American, and Middle Eastern or North African.

Covariates

We adjusted for covariates based on prior literature examining how anti-immigrant policies shape the take-up of social benefits, including age, gender, marital status, and children in the household. We also adjusted for total household income, education, and whether someone in the household received a stimulus check. We accounted for respondents’ awareness of the 2021 public charge rule's reversal with the following question, “How much have you heard about changes made to the public charge rule earlier this year by the Biden administration?” Response options were i) I heard a lot about it, ii) I heard something or a little bit about it, and iii) I did not hear anything about it. To assess knowledge of the 2021 public charge reversal, we asked respondents to evaluate the extent to which the following statement is correct: “Federal and state assistance with health care, food, or housing is only available for U.S. citizens, immigrants who are not citizens are not eligible for such help due to public charge.” The five response options ranged from “Definitely Correct” to “Definitely Incorrect.” At the time of the survey, Medicaid, SNAP, and housing vouchers were no longer considered when determining someone's public charge status, as the rule was reversed six months earlier. As such, we created three response categories: i) those who selected “Definitely correct or likely correct” were classified as having a limited understanding of public charge; ii) “Definitely incorrect or likely incorrect” were classified as having proficient understanding; iii) those selecting “Not sure if correct or incorrect” remained an independent category. These two questions were asked early in the survey to prevent endogeneity in the subsequent responses.

Statistical analysis

First, we examined the characteristics of the sample by respondents’ legal status and family composition. Differences in demographic and socioeconomic variables by legal status and family composition were determined using Pearson's chi-square test. We fit multivariable generalized linear models (family: binomial, link: logit) to measure the association between combinations of legal status and proximal cross-status ties and avoidance of public assistance programs due to immigration concerns. We used clustered standard errors at the state level (43 states plus the District of Columbia in our sample) to account for geographic variability in state-level immigrant policies that expand or restrict immigrants’ rights (Young et al., 2020). Generalized linear models provide a flexible framework for modeling outcomes that are not normally distributed and accommodate non-constant variance (Fox, 2015).

We used a stepwise approach for variable inclusion in the models. Model 1 examined the bivariate relationship between avoidance of needed assistance programs and combinations of legal status and proximal cross-status ties. Model 2 introduced a binary variable for ethnicity (Latino or Asian). Model 3 subsequently added whether respondents were aware of Biden's changes to the public charge rule in 2021 and an item assessing knowledge of what programs are considered in the current public charge rule. Model 4 introduced demographic factors as possible confounders of the association. Model 5 included social and economic factors as alternative explanatory variables for differences in avoidance of social assistance programs. After Model 5, we conducted Wald tests to check for joint significant differences between the coefficients for two LPI groups (with and without cross-status ties) and the reference group (Hypothesis 1a). We also assessed significant differences between LPI with and without cross-status ties (Hypothesis 1b) and between U.S. citizens with and without cross-status ties (Hypothesis 1c).

Model 6 included an interaction between respondents’ self-reported ethnicity and combinations of legal status and proximal cross-status ties to determine whether Asian and Latino individuals have distinct responses to anti-immigrant contexts (Hypothesis 2). We tested whether the overall interaction was significant. To provide ease of interpretation in disparities in avoidance of social assistance programs due to immigration concerns, we used marginal effects to estimate predicted probabilities, adjusting for multiple comparisons with Bonferroni-adjusted p-values. The variance inflation factor of the fully adjusted model was 2.02, indicating no multicollinearity. We completed all data cleaning, recoding, and analysis using Stata 17.0.

Sensitivity analysis

We combined the legally precarious immigrant groups (i.e., those with and without cross-status ties) to determine whether our results were sensitive to the differences in the family composition of these respondents. We also conducted multinomial logistic regression to examine whether findings were robust to the operationalization of the outcome variable as three distinct categories. We also conducted all regressions without weights to determine whether using weights artificially lowered statistical significance values. Finally, we accounted for respondents’ distal ties to noncitizens, including whether they had another family member, friend, or someone else who was a noncitizen, to determine whether the role of proximal family ties remained robust.

Results

Our analytical sample includes a total of 960 respondents. Table 1 presents the weighted sample characteristics. Overall, 21.3 % of the sample reported avoiding social assistance programs during the pandemic due to immigration-related concerns. A plurality of the sample was between the ages of 30 and 39 (31.2%), had a bachelor's degree (29.4%), had a household income of less than $30,000 (27.5%), and reported having children in the household (38.1%). A majority of the sample identified as female (52.0%, was married (53.9%), and reported they or a family member had received a stimulus check (75.9%).

Table 1.

Demographic characteristics by citizenship and family composition (n = 960).

	U.S. citizen + no cross-status ties	U.S. citizen + cross-status ties	LPR + cross-status ties	LPI + cross-status ties	LPI + no cross-status ties	Total	p-value
Demographics	N = 258	N = 401	N = 109	N = 141	N = 55	N = 960
Avoided needed assistance	17.5	21.8	20.2	21.1	36.2	21.3	0.12
Age group							0.001
18–29	36.4	24.3	17.8	26.2	27.0	27.0
30–39	31.2	26.8	35.9	35.2	41.0	31.2
40–49	16.0	22.4	20.7	22.5	19.3	20.4
50+	16.5	26.5	25.6	16.0	12.7	21.4
Sex at birth							0.025
Female	46.8	51.3	56.3	63.1	39.9	52
Male	52.8	48.2	42.7	36.9	60.1	47.6
Total household income							<0.001
≤$29,999	20.1	25	26.8	40.8	42	27.5
$30,000-$49,999	18.7	21.7	23.2	27.1	25.3	22.2
$50,000-$79,999	25	30.7	32.2	19.1	9.2	26.4
$80,000+	36.1	22.6	17.8	13	23.5	24.0
Marital status							0.13
Married	51.9	53.8	61.3	54.3	47	53.9
Living with partner, but not married	14.6	10.7	15.2	17.7	14.8	13.6
Not married	33.5	35.5	23.5	28	38.2	32.5
Respondent or someone in your household received a stimulus	87.2	84.2	70.6	54.0	38.9	75.9	<0.001
Children under 18 in the household	40.3	34	42.8	44.5	29.2	38.1	0.052
Positive information effect	50.8	52.1	39.1	47.0	48.4	49.1	0.097
Ethnicity							<0.001
Latinos	50.2	45.6	31.9	64.2	49.5	48.1
Asian	49.8	54.4	68.1	35.8	50.5	51.9
Education							<0.001
Some HS or HS degree	19.9	26.6	37.3	38.9	23.4	27.9
Some College	22.3	19.8	14.1	15.4	11.5	18.6
Associate's degree	9.9	15.2	11.6	16.4	2.1	12.8
At least bachelor's degree	47.9	38.4	37	29.3	63.0	40.7
Public Charge Knowledge							0.39
Limited	38.3	39.7	28.5	34.5	32.8	36.7
Unsure	27.7	26.0	36.2	34.6	33.0	29.5
Proficient	34.0	34.3	35.2	30.9	34.1	33.8
Awareness of public charge changes							<0.001
I heard a lot about it	29.7	21.9	16.6	12.8	15.0	21.4
I heard something or little bit about it	41.8	48.3	53.4	47.4	60.5	47.9
I did not hear anything about it	28.4	29.8	30.0	39.8	24.5	30.7

Note: LPR=lawful permanent resident, LPI=legally precarious immigrant, HS=high school.

Table 2 presents the multivariable regression results of the association between legal status/family composition combinations and avoidance of needed social assistance programs during the pandemic due to immigration-related concerns. In the crude bivariate model (Model 1), legally precarious immigrants (LPI) with cross-status ties (odds ratio (OR)=2.68, 95 % Confidence Intervals (CI): 1.55–4.64) and U.S. citizens with cross-status ties (OR=1.32, 95 % CI: 1.01–1.72) reported greater avoidance of social programs due to immigration concerns compared to U.S. citizens without cross-status ties. Model 2, which included ethnicity as a covariate, yielded results like those of Model 1. On average, Asian respondents were no different than Latino respondents in reported avoidance of benefits due to immigration concerns (OR=0.70, 95 % CI: 0.48–1.04). In Model 3, where we accounted for awareness of public charge rule changes and knowledge of the rule, U.S. citizens with cross-status ties (OR=1.64, 95 % CI: 1.18–2.29), LPI with cross-status ties (OR=1.94, 95 % CI: 1.26–2.98), and LPI without cross-status ties (aOR=4.0, 95 % CI: 2.12–7.56) were more likely to report avoidance of social assistance programs due to immigration concerns than the reference group. When we adjusted for demographic covariates, including age, gender, marital status, and children in the household in Model 4, the results remained consistent with those of the previous model. In Model 5, when adjusting for socioeconomic variables, the magnitude of the coefficients decreased slightly, but the significance and direction of the relationship endured for all groups. In the joint probability test, both LPI groups (with and without cross-status ties) were significantly more likely (chi2(2)=9.79, p-value=0.008) to report avoidance of social programs due to immigration concerns compared to U.S. citizens without familial cross-status ties (Hypothesis 1a). Further, LPI without cross-status ties were statistically more likely to have avoided social programs (chi2(1)=3.72, p-value=0.05) than their counterparts with ties to U.S. citizens with cross-status ties (Hypothesis 1b). We found that U.S. citizens with cross-status ties were significantly likelier to report avoidance of social programs due to immigration concerns (chi2(1)=7.12, p-value=0.0076) than their U.S. citizen counterparts without cross-status ties (Hypothesis 1c).

Table 2.

Avoidance of social assistance programs due to immigration-related concerns as a function of legal status and family composition, n = 960.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	odds ratio	odds ratio	odds ratio	odds ratio	odds ratio	odds ratio
Legal status + Family Composition (Ref. U.S. citizen + no cross-status ties)
U.S. citizen + cross-status ties	1.32*	1.34*	1.64⁎⁎	1.76⁎⁎	1.66⁎⁎	1.70*
	[1.01 - 1.72]	[1.03 - 1.74]	[1.18 - 2.29]	[1.22 - 2.54]	[1.14 - 2.40]	[1.08 - 2.68]
LPR + cross-status ties	1.20	1.28	1.74	1.82	1.70	0.42*
	[0.70 - 2.04]	[0.72 - 2.26]	[0.91 - 3.32]	[0.99 - 3.32]	[0.99 - 2.91]	[0.18 - 0.96]
LPI + cross-status ties	1.26	1.20	1.94⁎⁎	1.95⁎⁎	1.71*	1.33
	[0.84 - 1.89]	[0.82 - 1.76]	[1.26 - 2.98]	[1.30 - 2.92]	[1.14 - 2.57]	[0.79 - 2.26]
LPI + no cross-status ties	2.68⁎⁎⁎	2.70⁎⁎⁎	4.00⁎⁎⁎	4.28⁎⁎⁎	3.64⁎⁎	4.81⁎⁎
	[1.55 - 4.64]	[1.54 - 4.74]	[2.12 - 7.56]	[2.23 - 8.22]	[1.67 - 7.96]	[1.48 - 15.60]
Asian (Ref. Latinos)		0.70	0.87	0.93	0.97	0.68
		[0.48 - 1.04]	[0.59 - 1.29]	[0.62 - 1.39]	[0.69 - 1.35]	[0.38 - 1.22]
Awareness of public charge changes (Ref. Heard a lot about changes)
Heard a little about it			0.34⁎⁎⁎	0.37⁎⁎⁎	0.37⁎⁎⁎	0.36⁎⁎⁎
			[0.25 - 0.48]	[0.27 - 0.51]	[0.26 - 0.52]	[0.25 - 0.53]
Did not hear anything			0.13⁎⁎⁎	0.15⁎⁎⁎	0.15⁎⁎⁎	0.15⁎⁎⁎
			[0.08 - 0.22]	[0.09 - 0.25]	[0.08 - 0.26]	[0.08 - 0.25]
Knowledge of public charge (Ref. Limited knowledge)
Unsure			0.46⁎⁎⁎	0.44⁎⁎⁎	0.42⁎⁎⁎	0.40⁎⁎⁎
			[0.29 - 0.72]	[0.30 - 0.66]	[0.28 - 0.64]	[0.25 - 0.65]
Satisfactory knowledge			0.46*	0.47*	0.45⁎⁎	0.44⁎⁎
			[0.25 - 0.83]	[0.26 - 0.86]	[0.25 - 0.82]	[0.24 - 0.82]
Age group (Ref. 18–29)
30–39				0.72	0.70	0.67
				[0.45 - 1.14]	[0.45 - 1.07]	[0.44 - 1.03]
40–49				0.91	0.89	0.85
				[0.53 - 1.56]	[0.53 - 1.50]	[0.51 - 1.42]
50+				0.57⁎⁎	0.51⁎⁎⁎	0.51⁎⁎⁎
				[0.38 - 0.83]	[0.35 - 0.75]	[0.35 - 0.74]
Marital Status (Ref. Married)
Living with partner, but not married				1.00	1.03	0.96
				[0.54 - 1.84]	[0.55 - 1.93]	[0.54 - 1.73]
Not married				0.91	0.91	0.89
				[0.60 - 1.40]	[0.59 - 1.42]	[0.58 - 1.37]
Child in household (Ref. Yes)
No children in household				0.63*	0.61*	0.63*
				[0.43 - 0.93]	[0.41 - 0.92]	[0.43 - 0.93]
Gender (Ref. Female)
Male				1.14	1.15	1.15
				[0.88 - 1.48]	[0.86 - 1.55]	[0.88 - 1.49]
Total household income (Ref. <$30,000)
$30,000-$49,999					0.52⁎⁎⁎	0.50⁎⁎⁎
					[0.36 - 0.74]	[0.35 - 0.71]
$50,000-$79,999					0.76	0.74
					[0.55 - 1.05]	[0.52 - 1.04]
$80,000+					0.41⁎⁎⁎	0.41⁎⁎⁎
					[0.27 - 0.63]	[0.26 - 0.66]
Respondent or household member received stimulus (Ref. Yes)
Did not receive stimulus					1.03	0.97
					[0.72 - 1.49]	[0.65 - 1.46]
Educational attainment (Ref. At least some HS or HS graduate)
Some College					0.56	0.54
					[0.30 - 1.04]	[0.28 - 1.01]
Assoiciate's degree					1.49*	1.45
					[1.01 - 2.20]	[0.97 - 2.16]
At least bachelor's degree					1.47	1.47
					[0.92 - 2.35]	[0.91 - 2.36]
Ethnicity X Legal Status/Family Composition (Ref. Asian + U.S. citizen no cross-status ties)
Asian#U.S. citizen + Cross-status ties						1.07
						[0.61 - 1.86]
Asian# LPR + Cross-status ties						8.76⁎⁎⁎
						[3.09 - 24.87]
Asian# LPI + Cross-status ties						2.18
						[0.98 - 4.84]
Asian#LPI + no cross-status ties						0.67
						[0.14 - 3.18]
Constant	0.21⁎⁎⁎	0.25⁎⁎⁎	0.76	1.04	1.55	1.93
	[0.16 - 0.29]	[0.18 - 0.35]	[0.47 - 1.24]	[0.55 - 1.96]	[0.62 - 3.90]	[0.78 - 4.75]

^⁎⁎⁎p < 0.001,.

^⁎⁎p < 0.01,.

^⁎p < 0.05, + p < 0.1

Note: LPR=lawful permanent resident, LPI=legally precarious immigrant, HS=high school. Clustered standard errors at state-level in brackets.

Finally, in Model 6, the interaction between ethnicity and legal status/family composition combinations was statistically significant, p = 0.001, chi2(4) = 27.1. This interaction represents the second partial derivative with respect to both ethnicity and legal status/family composition. Only the coefficient for Asian LPR was significant, suggesting that ethnicity only had a moderating association with legal status/family composition among LPRs. That is, Asian LPRs had greater odds of avoiding social assistance programs due to immigration concerns than Latino LPRs, all relative to the difference between Asian and Latino U.S. citizens without cross-status ties. We confirmed this interaction with a joint test of the effects of legal status and family composition within each level of ethnicity (Latinos vs. Asians). We found one statistically significant joint test: compared to Latino LPRs with cross-status ties, Asian LPRs with cross-status ties had an additional 21.9 percentage-point increase in the probability of avoiding social assistance programs due to immigration concerns (Bonferroni adjusted p-value = 0.024). These predicted probabilities are displayed in Fig. 1.

Fig. 1.

Predicted probabilities of avoiding social assistance due to immigration among Latinos and Asian respondents (n = 960) based on Model 6 from Table 2. Note: LPR = lawful permanent resident, LPI = legally precarious immigrant.

Table 3 presents the predicted probabilities of differences in forgoing needed social assistance due to immigration concerns from the average marginal effect models. All marginal effect models control for demographics, socioeconomic characteristics, and awareness and knowledge of the 2021 public charge rule reversal. In Panel A, we present the predicted probabilities (PP) corresponding to Table 2 Model 5, in which we quantified the average differences in the likelihood of avoidance of social assistance programs due to immigration concerns across legal status and family composition among all respondents. Approximately 15.69 % (95 % CI: 12.0–19.39) of U.S. citizens without cross-status ties reported they or a family member did not use the social assistance programs they needed due to immigration concerns. Except for LPR with cross-status ties, all other groups experienced a significantly higher probability of avoiding social assistance programs due to immigration concerns. The highest percentage was reported by 35.19 % of legally precarious immigrants without cross-status ties. Panel B presents the PP of avoiding social programs due to immigration concerns based on the interaction between ethnicity and legal status/family composition combinations (Table 2 Model 6). We found that Latino U.S. citizens without cross-status ties had a 17.26 % (95 % CI: 11.94–22.59) probability of avoiding social programs due to immigration concerns. Only legally precarious Latino immigrants without cross-status ties (PP: 43.02 %, 95 % CI: 22.63–63.42) had a significantly higher probability of avoiding social assistance programs than Latino U.S. citizens without cross-status ties. Among Asian respondents, we found that only LPRs with cross-status ties (PP: 30.82 %, 95 % CI: 23.79–37.84) were more likely than U.S. citizen Asians without cross-status ties (PP: 13.05 %, 95 % CI: 8.49–17.61) to have a higher probability of avoiding social assistance programs due to immigration concerns. Finally, Asian LPRs with cross-status ties had a 21.9 percentage point higher probability of avoiding social assistance programs due to immigration concerns compared to Latino LPRs (8.92 % vs. 30.82 %), as indicated in the interaction between ethnicity and legal status/family composition.

Table 3.

Predicted probabilities of avoiding social assistance due to immigration concerns among Latinos and Asian respondents (n = 960).

Panel A: Average marginal effects across all respondents(corresponds toTable 2Model 5)
All (n = 960)	Predicted Probability	95 % Confidence Interval
U.S. citizens + no cross-status ties	15.69 %	12.00 %	19.39 %
U.S. citizen + cross-status ties	22.17 %*	19.40 %	25.17 %
LPR + cross-status ties	22.53 %	16.81 %	28.25 %
LPI + cross-status ties	22.62 %*	19.39 %	26.10 %
LPI + no cross-status ties	35.19 %*	21.81 %	48.57 %

Panel B: Average marginal effects from the interaction between ethnicity and legal status/family composition(corresponds toTable 2Model 6)
Latinos (n=499)	Predicted Probability	95 % Confidence Interval
U.S. citizens + no cross-status ties	17.26 %	11.94 %	22.59 %
U.S. citizen + cross-status ties	24.50 %	18.80 %	30.21 %
LPR + cross-status ties	8.92 %^	2.02 %	15.82 %
LPI + cross-status ties	20.98 %	16.25 %	25.71 %
LPI + no cross-status ties	43.02 %*	22.63 %	63.42 %
Asian (n=461)	Predicted Probability	95 % Confidence Interval
U.S. citizens + no cross-status ties	13.05 %	8.49 %	17.61 %
U.S. citizen + cross-status ties	19.91 %	16.92 %	22.91 %
LPR + cross-status ties	30.82 %*^	23.79 %	37.84 %
LPI + cross-status ties	26.91 %*	20.34 %	33.47 %
LPI + no cross-status ties	28.53 %	10.91 %	46.14 %

Average marginal effect models in Panel A controls for ethnicity, awareness of public charge, age, income, marital status, children in the household, receipt of pandemic stimulus. Average marginal effect models in Panel B controls for all variables controlled in Panel A except for ethnicity, which is an interaction term.

^⁎Corresponds to a significant difference compared to the all-U.S. citizen households group within each ethnic group (i.e., comparing U.S. citizens with no cross-status ties to other legal status/family composition categories within each ethnic group).

^{^}Corresponds to a significant joint test of the effects of legal status and family composition across each level of ethnicity (i.e., comparing Latino and Asian respondents at each level of legal status/family composition categories).

Note: LPR=lawful permanent resident, LPI=legally precarious immigrant.

Sensitivity analyses

We found consistent results in the models in which we combined all legally precarious immigrant respondents regardless of family composition (see Appendix Table A). Legally precarious immigrants, irrespective of cross-status ties, were more likely than U.S. citizens without cross-status ties to report avoiding social assistance programs due to immigration concerns in the fully adjusted model (aOR=2.49, 95 % CI: 1.7–3.7), albeit at a lower magnitude than legally precarious immigrants in all-noncitizen households in Table 2 Model 5 of the primary analysis (aOR=3.64, 95 % CI: 1.67–7.96). Our confidence intervals were narrower in the sensitivity analyses since we had a larger cell size of LPI.

In the fully adjusted multinomial logistic regressions, where we compared those who a) needed and avoided social programs, b) reported no need for assistance, and c) applied and received assistance (reference), we found consistent results. Compared to U.S. citizens without cross-status ties, U.S. citizens, LPR, and legally precarious immigrants with cross-status ties and legally precarious immigrants without cross-status ties were more likely to have reported avoiding assistance programs than having applied and received assistance (see Appendix Table B). The magnitude of the coefficient was most substantial for legally precarious immigrants without cross-status ties. Second, compared to U.S. citizens without cross-status ties, LPR with cross-status ties were the only group to report that they did not need assistance relative to those who applied and received assistance. Our results remained robust in the unweighted analyses (Appendix Table C) and in the analysis where we adjusted for distal ties to noncitizens (Appendix Table D).

Discussion

We tested the association between combinations of legal statuses and family composition, that is, the presence or absence of cross-status family ties, and avoidance of social assistance programs due to immigration concerns during the later stages of the COVID-19 pandemic when the public charge rule was undergoing a reversal. Our data allowed us to describe population patterns of Asian and Latino groups’ experiences and assess whether legal status and family composition had a distinct influence on either group's decisions to not use needed social programs due to immigration-related concerns.

Consistent with qualitative data on mixed-status families’ experiences during the pandemic, we found that across the sample, legally precarious immigrants with and without cross-status ties were more likely to report that they avoided social programs due to fear of immigration consequences than U.S. citizens without cross-status ties (supporting Hypothesis 1A). Legally precarious immigrants living in households where everyone was a noncitizen, that is, without cross-status ties, consistently had the highest probability of reporting avoiding the social assistance programs they needed due to immigration concerns regardless of ethnicity.

Further, we acknowledge that immigrants’ experiences with immigration policies are also a function of their family composition. For instance, despite the small sample size, legally precarious individuals in a household where all family members were non-citizens, that is, lacking cross-status ties, were the only group that consistently was more likely to report avoiding social assistance due to immigration concerns. They were also more likely than their counterparts with cross-status ties to report avoiding social programs (supporting Hypothesis 1B). This finding may suggest that living with U.S. citizens in one's household may offer some, albeit limited, protection to legally precarious family members. At the same time, the adverse effects of anti-immigrant policies have spillover effects on individuals with U.S. citizenship. We found that U.S. citizen respondents with cross-status ties to a noncitizen were likelier to report avoiding social assistance due to immigration concerns than U.S. citizens without cross-status ties (supporting Hypothesis 1C).

Our findings highlight that not all immigrants experience exclusionary policies similarly. We find significant differences in who avoided needed social assistance programs between Asian and Latino individuals across legal status and family composition (supporting Hypothesis 2). The interaction between ethnicity and legal status/family composition suggests nuances in how people with lawful permanent residence (LPR) or green card holders experience the public charge rule. Asian LPRs with cross-status ties were more likely to report that they or a family member avoided social assistance compared to their Latino counterparts. This finding was obscured in the main effects models, where we combined all respondents across ethnicity and found no significant differences in the likelihood of reporting avoiding social assistance due to immigration concerns between LPR and U.S. citizens in all-citizen households. Given that public charge does not apply to people with LPR status, this finding highlights the importance of combatting misinformation regarding the policy among this group.

What may explain this pattern? Asian individuals may have fewer organizational resources or networks to learn about public charge updates. For example, past studies have found that a majority of nonprofit organizations supporting undocumented organizations are primarily Latino-serving (Cho, 2017). Further, even with the availability of pan-ethnic organizations serving all immigrants, Asian individuals may be less likely to access the resources available. For instance, previous studies have found that young Asian undocumented immigrants evaded surveillance and avoided accessing resources for undocumented students (Enriquez, 2019). It is also possible that Asian individuals did not have the financial need to apply for public benefits. In our sample, for instance, Asian respondents were more likely to have higher incomes than their Latino counterparts.

It is crucial for organizations that serve immigrants to establish trust with diverse pan-ethnic communities to combat misinformation and address concerns regarding public charge. In March 2021, the public charge expansion was reversed to the 1999 Guidelines. However, the various changes, injunctions, and lawsuits following the Trump Administration's changes to public charge, along with ambiguous implementation guidance of the latest policy from the Biden Administration, may have caused confusion among immigrants and their families. Combating misinformation about what constitutes a public charge requires governmental resources to ensure that immigrants feel comfortable using resources they and their families qualify for without fear of future immigration-related consequences (Chaparro et al., 2023). Supportive public institutions, including immigrant rights nonprofits, can focus on clear communication and outreach to marginalized Asian and Latino communities to help combat misinformation regarding public charge.

An essential aspect of this study is the importance of assessing immigrants’ awareness and knowledge of policies that may affect them when evaluating the policies’ consequences. For instance, in our analysis, before adjusting for awareness of the public charge rule changes and knowledge of the social programs considered in public charge cases, legal status/family composition differences were only evident among legally precarious immigrant respondents in all-noncitizen households. Once we accounted for awareness and knowledge of the policy, legally precarious respondents with cross-status ties and U.S. citizens with cross-status ties were more likely than the reference group to report avoiding social assistance due to immigration concerns. Studies that estimate the potential number of people who could be affected by the public charge rule may produce biased results on who is avoiding social programs, given the heterogeneity in awareness of the policy change and confusion as to what programs count toward future inadmissibility.

Strengths and limitations

Our study has limitations worth noting. First, the cross-sectional nature of our data preempts any causal claims. Nonetheless, these findings serve as preliminary evidence of the legal and family composition disparities in avoiding social assistance due to immigration concerns during a national public health emergency. Second, we have a small sample size of legally precarious immigrants, especially in all-noncitizen families. Despite this, our findings for LPI in all-noncitizen families were consistent and significant across every model and in the sensitivity analyses when we combined the undocumented categories across their family composition. Nonetheless, imprecise estimate effects and wide confidence intervals resulting from the interaction may prevent us from finding other differences between Asian and Latino respondents. Studies with a larger sample size of ethnically diverse LPI and undocumented respondents could help uncover differences in the patterning of behaviors and experiences resulting from immigration policies. Third, we do not ask directly about the social programs, such as Medicaid, SNAP, or housing vouchers, that respondents avoided. Future work can disentangle the use of these programs to inform interventions and informational campaigns to combat fear and confusion regarding eligibility. Fourth, self-selection may introduce potential non-randomness in the survey distribution, impacting the representativeness of our sample. However, as our primary focus is on analyzing the relationship between cross-status ties and the avoidance of social policy programs due to immigration fears, rather than estimating population parameters, the potential bias does not critically undermine our conclusions. Finally, it is possible that state-level context matters, including restrictive and immigrant-friendly policies, in shaping the extent to which immigrants avoid programs based on immigration fears. Future research should account for state-level policies that set the stage for implementing federal policy changes when examining how social assistance avoidance due to immigration concerns influenced immigrants’ access to healthcare during the pandemic, including the uptake of the COVID-19 vaccine. Lastly, we do not assess the perceived social support that familial cross-ties provide. Future studies should collect data on noncitizen persons’ perceptions of the support U.S. citizen family members provide and also assess whether distal ties to U.S. citizens offer any support.

Despite the limitations, this study contributes to the growing literature on how policies affect the well-being of immigrants and their families. We examined the differences in legal status and family composition in unmet social assistance needs due to immigration-related concerns. While previous studies experienced methodological challenges ascertaining which individuals have a precarious legal status, we have detailed information on respondents’ legal status, including whether they were DACA recipients or had another liminal status. By disaggregating noncitizens (green card holders and respondents with precarious legal status), we gained insight into the patterning of a diverse group of immigrants’ experiences during a period marked by a global pandemic and heightened anti-immigrant policies.

Conclusion

We found that illegality spills over to affect family members who are not the direct target of exclusionary policies like public charge. For instance, U.S. citizens with noncitizen family members, compared to those in all-U.S. citizen households, had a greater probability of avoiding social programs due to immigration fears. Similarly, legally precarious immigrants without cross-status ties to a U.S. citizen family member, that is, where all household members were noncitizens, were consistently more likely to avoid social programs they needed due to immigration concerns compared to their counterparts with cross-status ties. Finally, addressing misinformation regarding the use of public assistance programs and immigration consequences among Asian and Latino communities could prevent the widening of health inequities.

Ethical statement

Hereby, I, Alein Haro-Ramos, consciously assure that for the manuscript, “Immigration Concerns and Social Program Avoidance: The Roles of Legal Status and Familial Cross-Status Ties among Asian and Latino Communities,” the following is fulfilled:

1) This material is the authors' own original work, which has not been previously published elsewhere.

2) The paper is not currently being considered for publication elsewhere.

3) The paper reflects the authors' own research and analysis in a truthful and complete manner.

4) The paper properly credits the meaningful contributions of co-authors and co-researchers.

5) The results are appropriately placed in the context of prior and existing research.

6) All sources used are properly disclosed (correct citation). Literally copying of text must be indicated as such by using quotation marks and giving proper reference.

7) All authors have been personally and actively involved in substantial work leading to the paper and will take public responsibility for its content.

I agree with the above statements and declare that this submission follows the policies as outlined in the Guide for Authors and in the Ethical Statement.

CRediT authorship contribution statement

Alein Y. Haro-Ramos: Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization, Project administration, Writing – review & editing. Gabriel Sanchez: Investigation, Data curation. Matt Barreto: Resources, Data curation.

Declaration of competing interest

The authors have no conflict of interest or financial disclosures to release.