Beyond undocumented: Differences in the mental health of Latinx undocumented college students

Undocumented college students face several threats to their well-being and mental health. Different social locations, including whether students have Deferred Action for Childhood Arrival (DACA) status, students’ gender, and family factors may shape students’ ability to be well. How these factors work together to shape mental health outcomes among undocumented Latinx college students is not well understood. This study examines several factors (demographic, familial, immigration, and socioeconomic factors) associated with anxiety scores of undocumented Latinx college students who participated in the UndocuScholars Project national online survey in 2014. We observe three notable findings: (1) DACA recipients report heightened levels of anxiety, (2) women with DACA status report higher levels of anxiety compared to non-DACAmented undocumented college students and men with DACA, and (3) students whose families motivate them report lower levels of anxiety. Latinx undocumented college students are not a monolith; demographic, family, and socioeconomic factors matter.


Palabras clave Latino · Inmigrantes indocumentados · DACA · Salud mental
Ever since September of 2017, when the Trump administration announced their intention to terminate the Deferred Action for Childhood Arrivals (DACA) program, DACA has been threatened time and time again. In July 2020, the US Supreme Court ruled against the Trump administration's decision to suspend DACA. After numerous legal battles, the DACA program has been fully reinstated (and is accepting new applications as well as renewals as of 7 December 2020). The legal battle for DACA, despite its reinstatement, continues (e.g., Texas and several other states filed a lawsuit against the federal government over the legality of DACA; this issue was heard by the Texas federal court in December 2020, but there was no immediate ruling). This series of political moments displays, with much clarity, the uncertain nature of DACA. It reminded recipients and their supporters that DACA was never a permanent way to protect undocumented youth and young adults. The wellbeing of DACA recipients was likely dampened by these political moments (Venkataramani et al. 2017;Enriquez et al. 2018;Hamilton et al. 2020), but to date, we lack a clear understanding of both individual-and family-level factors that protect or worsen the mental health of DACA and non-DACA recipients. In the present study, we examine this issue in a sample of Latinx 1 undocumented undergraduates using survey data collected in 2014. Our study examines mental health in a period during which DACA was arguably more "stable" than it is today. Our findings reveal that, even during 2014, anxiety levels of undocumented Latinx undergraduates with and without DACA were troubling and varied on the basis of demographic, familial, and socioeconomic factors.
Across the United States, states have passed laws that reduce financial barriers to attend college by implementing in-state tuition for, and/or providing financial aid to undocumented students (NCLS 2019). Each year, 65,000 undocumented youth graduate from high school nationwide, and about 5% to 10% pursue higher education (Campaign for College Opportunity 2018). More than 450,000 undocumented individuals are enrolled in higher education in the United States (New American Economy 2020). Latinx individuals constitute 46% of the undocumented undergraduate population, and among DACA-eligible undergraduates, 65% were Latinx in 2018 (New American Economy 2020).
Despite state policies that facilitate college access, several other factors influence whether Latinx undocumented college students thrive. Though there is limited national-level data about the anxiety levels of Latinx youth and undocumented college students, previous studies in smaller regions find troubling results. Immigrant Latinx youth in North Carolina are at higher risk of having anxiety (28%) compared with US-born youth (13-20%) (Potochnick and Perreira 2010). A report from the UndocuScholars Project survey, the online survey of undocumented college students used in this study, found that about 29% of men and 37% of women in the Undo-cuScholars Project survey reported anxiety levels above the clinical cutoff (Teranishi et al. 2015). Within undocumented individuals, DACA-eligible adults tend to report lower levels of psychological distress compared to non-DACA eligible immigrants (Venkataramani et al. 2017).
A combination of risk and protective factors shape the mental health profiles of Latinx youth (Potochnick and Perreira 2010). DACA is an especially intriguing factor. Patler and Laster Pirtle (2018) found that DACA recipients and non-DACA recipients worry about family deportation at comparable levels and that DACA recipients have slightly fewer worries about their own deportation. Enriquez and colleagues (2019) found that undocumented undergraduates attending colleges in the University of California system are extremely worried about DACA's future. DACA's uncertainty, the continued risk of family members' deportation, and the effects of anti-immigrant sentiment may undermine the mental well-being of all undocumented individuals (Dreby and Stutz 2012;Gonzales et al. 2013), as well as undermine the promise that DACA was previously thought to have had (Hamilton et al. 2020).
It is important to understand the dimensions that shape the mental health of Latinx undocumented students for several reasons. 60% of the approximately 125,000 undocumented students who graduate high school every year are Latinx (Zong and Batalova 2019), and one in twenty US-born children have an undocumented parent (Passel et al. 2018). A substantial number of young Latinx persons have ties to the undocumented community either through direct family and/or community ties (Vargas et al. 2017). Regardless of documentation status, Latinx identity is associated with stereotypes and tropes about the undocumented population. Because of this racialized illegality, membership in the Latinx and undocumented populations becomes conflated in public discourse and by the media (Menjívar 2021;Enriquez et al. 2019;García 2017). This homogenizes the Latinx experience. Thus, our study aims to challenge monolithic portrayals of the Latinx experience and of the undocumented experience (Enriquez et al. 2018(Enriquez et al. , 2019Valdez and Golash-Boza 2018). Lastly, mental health has implications for retention in higher education. One study examined Latinx college students' cognitive disruption pre-and post-Trump's election and found that when students were prompted to think about their familial obligations post-Trump, they displayed attentional disruption, indicating that their ability to focus on school was compromised (Vasquez-Salgado et al. 2018). Latinx mental health is a matter of retention in higher education, which influences social mobility (Wyatt et al. 2017).
No study to date has examined how demographic, family, immigration, and socioeconomic factors work together to create distinct mental health experiences among undocumented Latinx undergraduates with and without DACA. Our findings add unique support for the argument that DACA does not eliminate angst related to immigration (Hamilton et al. 2020). We rely on unique data from a national online survey with measures on worries about family deportation and DACA status. We argue that the mental health of undocumented undergraduates is complex and that demographic, family, immigration, and socioeconomic factors create diverging experiences.

Mental health and DACAmented status
The Obama administration implemented the DACA program in 2012 as an executive order. This program provided eligible undocumented individuals permission to legally work in the United States and deemed them low priority for deportation. Eligible individuals received a work authorization card and a social security number, and gained access to resources that could improve their social and economic incorporation: they could apply for driver's licenses, gain health insurance, develop credit, and apply for Advanced Parole (Zhou and Gonzales 2019;Gonzales et al. 2014). If granted Advanced Parole, DACA recipients may have the opportunity to deepen relationships with family in their home countries if traveling for humanitarian, emergency, or educational reasons (Ruth et al. 2019). In states such as Connecticut and Maryland, having DACA means having access to financial aid (NCLS 2019). More than 800,000 young adults have benefited from DACA (Zong and Batalova 2019). Individuals who did not meet all the criteria, however, could not benefit from the program. Because of the Trump administration's temporary suspension of DACA in September 2017 and other changes during 2020, the DACA program shortly changed from a two-year renewal work authorization to a one-year program. Since DACA's reinstatement in its original form on 7 December 2020, DACA-eligible individuals can now apply for two-year employment authorizations and Advanced Parole again.
DACA once held the promise of a brighter future for young undocumented adults, but it has had positive effects (Lee 2018;Lim 2018) as well as unintended effects (Hsin and Ortega 2018). DACA relieved individuals of stressors associated with the lack of a social security number, reduced feelings of shame as well as isolation (Patler and Laster Pirtle 2018), and increased individuals' sense of belonging (Siemons et al. 2017). But, it also prompted undergraduates in four-year colleges to make difficult choices between engaging in the labor market full time or leaving college (Hsin and Ortega 2018). DACA, thus, may have competing influences on mental health.

Stress process theory
Stress process theory is useful for understanding risk and protective factors influencing mental health (Pearlin 1989). This theory posits that social status influences exposure to stressors, defined as any event that "challenges the adaptive capabilities of people" (Pearlin 2010, p. 208). Stressors may be acute or long-term/chronic. One stressor may lead to others (i.e., low-income status may lead to having more family responsibilities). Protective factors represent the resources individuals have at their disposal to mitigate the harmful effects of stressors; these include social supports, coping mechanisms, and beliefs. This study focuses on one of the multiple dimensions of mental health that stressors impinge upon, self-reported levels of anxiety (Pearlin 2010).
The stress process framework is useful for understanding how long-term stressors, such as the threat of deportation, influence undocumented Latinx college students (Vargas et al. 2019;Dreby 2012). Some individuals may have more resources at their disposal to mitigate the impact of stressors on mental health. In this study we examine a unique combination of factors that protect or dampen mental health.

Family deportation worries
The risk of family deportation may increase anxiety among DACAmented and non-DACAmented college students. Although DACA beneficiaries are less concerned about their own deportation, they still worry about their family's deportation Patler and Laster Pirtle 2018;Childs 2018;Castañeda and Melo 2014;Dreby 2015). As stated by Golash-Boza and Valdez, "the fact that their family members are not safe is never far from their minds" (2018, p. 546). Some family relationships and ties may worsen the mental health of undocumented individuals (Del Real 2018;Vargas et al. 2017).
The extent to which one worries about their family's deportation may influence whether DACAmented individuals have better mental health compared to individuals without DACA. If DACAmented students continue to heavily worry about their family's deportation, positive effects of DACA on mental health may disappear, making mental health levels between DACA and non-DACA recipients more similar.

Family as motivation and support
Latinx college students' perceptions about their families may matter for mental health. Emotional connection to one's family is associated with well-being among Latinx students (Gándara 1995;Hurtado et al. 1996;Rodriguez et al. 2003;Sánchez et al. 2005;Solberg and Villarreal 1997;Sy and Romero 2008).
Latinx undergraduates who are the first in their family to attend college are often proud about being a first-generation college student. One's family-based motivation to continue school promotes psychological well-being (Mount 2015). In promoting positive mental health outcomes, family-based motivation or ganas (a strong desire to overcome) may mitigate the effect of stressors on anxiety, especially stressors related to being undocumented (Allen et al. 2020;Punti 2018). An undocumented college student who is deeply motivated by her family to go to college may have better mental health than a student whose family does not represent a high motivational purpose, because family may be a positive force in the educational journeys of students of color (Yosso 2005). At the same time, strong family-driven desires to succeed may coexist with family demands that curtail educational journeys (Punti 2018).
Family has been relatively understudied in analyses of undocumented college students' anxiety. Previous studies on young Latinx undocumented undergraduates have focused primarily on social network supports outside of the home (such as peers in school and institutional gatekeepers) (Kam et al. 2020(Kam et al. , 2019Patler 2018). One study of Latina undocumented undergraduates in a rural town found that, in response to negative encounters with college personnel, undocumented college students feel isolated and do not feel supported by their college campus (Muñoz 2013). As a coping mechanism, students isolate themselves and keep silent because they are afraid to speak with counselors who may not understand them (Muñoz 2013). Undocumented students who want to minimize risk and keep their family's undocumented status a secret may choose to conceal their status to friends or teachers (Kam et al. 2019;Patler 2018). Using isolation and silence as coping mechanisms may make undocumented undergraduates rely on their families more if they find college campuses alienating (Muñoz 2013). If students are closer to their families because they feel they are around people who understand them, family may be an important source of support for students (Cobb et al. 2016). However, we know relatively less about how family-level sources of support may mitigate negative mental health outcomes.
Having positive perceptions about one's family may be protective for student mental health. Family members raise the academic aspirations of immigrant youth and are a source of motivation to continue in their educational journeys Tienda 1995, 1998;Katsiaficas 2015;O'Neil et al. 2016;Portes 2010). Pérez and colleagues (2010) found that undocumented undergraduates make sense of their educational journeys by referring to how their parents encouraged them to pursue college. Latinx immigrant students are often motivated by their immigrant families because they want to "repay" them for all their sacrifices (Alcántara 2018;Jabbar et al. 2019).
In addition to motivation, family may provide tangible resources that promote Latinx undergraduates' mental health. Family may provide support ranging from financial help for school, rent, and bills to educational advice. One study found that first-generation Latinx college students in Texas received financial assistance for tuition, rent, or bills from their families (Jabbar et al. 2019).

Gender and mental health
Gender, which we define as a binary variable because of the data limitations, may matter for how one internalizes their immigration status because gender shapes how individuals report well-being. Previous regional studies have found that Latinx college students may be at heightened risk of anxiety symptoms compared with non-Latinx young adults in college (Zvolensky et al. 2019), and that there is variation by gender. For instance, one study found that the percent of Latinx undocumented college students in California with anxiety levels above the clinical cutoff levels is 25% for men and 35% for women (Suárez-Orozco and López Hernández 2020). To place these numbers in context, recent estimates suggest this figure is about 31% among the general college student population (American College Health Association 2020). In the broader population, the prevalence of any severity of anxiety is about 19% among women and 11.9% among men (Terlizzi and Villarroel 2020).
Beyond gender differences in reporting mental health outcomes, Longest and Thoits (2012) found that the number of risk factors that men need to experience to express high levels of distress are very high compared with those of women. At the same time, the deportation regime is gendered, as deportations disproportionately affect men of color (Golash-Boza 2015). If undocumented Latinx men are aware of this surveillance, they may experience heightened anxiety levels. Whether they report this on surveys is difficult to detect, however. We examine one aspect of mental health using self-reported anxiety, and we keep in mind that gender ideologies, expectations, and scripts influence reports of anxiety (Longest and Thoits 2012;Hill and Needham 2013).

Gender, family responsibilities, and mental health
Gender may shape Latinx college students' mental health through familial responsibilities. Gender ideology shapes the social experiences of undocumented migrants (Donato et al. 2017;Enriquez 2017), and familial responsibilities of children are gendered from an early age (Orellana 2001;Quiroz-Becerra 2013, p. 147). In addition to schoolwork, Latina immigrant children disproportionately help with cleaning, cooking, caretaking, and translating for kin (Valenzuela 1999;Orellana et al. 2003;Estrada and Hondagneu-Sotelo 2013, p. 145). Socioeconomic status shapes immigrant children's participation in the work of their parents (Estrada and Hondagneu-Sotelo 2013, p. 145). Among Latinx undocumented college students, gendered norms may create differences in the amount and type of familial responsibilities women and men engage in.
According to Estrada, children's "own labor contributions are what make it possible for their families to survive the structural economic and employment barriers they face in the lower sector of the economy" (Estrada 2019, p. 16). These contributions may persist in undocumented college students, but we know little about the mental health consequences of familial responsibilities among DACAmented and non-DACAmented Latinx undergraduates. Supporting family has varied consequences. Vallejo found that Mexican middle-class individuals in Southern California who grew up economically disadvantaged provided financial support to kin during college and through adulthood, but these responsibilities at times hindered social mobility (Vallejo 2012).
Limited studies focus on gender and family factors among undocumented young adults. Enriquez (2017) found that undocumented men stall their family formation because they feel they cannot fulfill expectations to be caretakers of their own family. Pressures and expectations remain gendered within the undocumented community, and social life is not gender-neutral for undocumented individuals.
If women follow hegemonic gender roles, then Latinas in the study might carry heavy burdens of worry and familial responsibilities in addition to college-related stressors. Women with DACA may provide for their families in ways they may not have been pressured to before having DACA. If women take on more labor for their families, they might have heightened anxiety scores. Undocumented Latino men may have gendered expectations to provide for their families regardless of DACA status; therefore, their expressions of mental health may not be as affected by DACA status.

Research questions of the present study
This study examines protective and risk factors that influence mental health, measured by self-reported anxiety, among Latinx undocumented undergraduates who participated in the UndocuScholars Project online survey. Our research questions are 1. What is the relationship between protective/coping factors (DACA, family motivation, family support) and risk factors (family deportation worries, gender, and familial responsibilities) and mental health?
a. Do DACA recipients have better mental health compared with non-DACA recipients? Do worries of family deportation reduce the mental health of DACA recipients? b. Given the large literature on gender and family factors, do women with DACA face similar mental health profiles as men?

Data and variables in the analysis
This study relies on data from the UndocuScholars Project, a study created in response to the need for research on undocumented college students (Teranishi et al. 2015). Because of their stigmatization, invisibility, and lack of available data, undocumented undergraduates are a "hard-to-reach" population (Marpsat and Razafindratsima 2010). Thus, novel strategies were employed to reach the sample. The primary method of recruitment of participants for this project was through a web portal and a strong, multi-platform social media campaign (e.g., Facebook, Twitter, Instagram) that focused on providing information about undocumented issues and provided the opportunity to have student voices heard by participating in the study. Additionally, participants were recruited by partnering with organizations that worked with undocumented students. Lastly, participants were recruited through flyers, announcements at college events, and word of mouth.
To take the survey, participants must have met these eligibility criteria: (1) Be between eighteen and thirty years of age; (2) identify as an undocumented, DREAMer, or DACAmented college student; and (3) have been enrolled as an undergraduate student in college or university in the past year. Because of the nature of the topic, participants were assured anonymity. Consent for participation was a simple checkmark before starting the survey. For the online survey, we did not store any identifiable information such as IP addresses. A data control protocol was implemented to reduce the number of mischievous survey data-for example, we flagged responses: when only a limited amount of time (< 10 min) was spent completing the survey, when there was a mismatch between language spoken at home and country of origin, and cases in which qualitative responses were in verbatim to others. When any of these occurred, the flagged surveys were checked by a group of research team members. Once the survey was deemed valid, each participant received a $20 gift card in return for their participation. The survey was made in Qualtrics within the project's website. To ensure anonymity, once a survey response was deemed legitimate, participants were sent a link with their gift card, and their email, the only information linking their responses to a personal information item, was deleted from the server (Teranishi et al. 2015). The total survey sample was 909 students, of which 807 identified as Latinx. Once missingness of key variables was taken into account, 660 Latinx college students remained in the analytic sample.
This study brought together experts from academia and the community. The UndocuScholars Project team included a student advisory board and a community advisory board. The second author of this article was a part of the UndocuScholars Project team. The first author was not part of the data collection team. Both authors are Latinx women, and the first author is an immigrant with a similar background as that of the study participants. Both authors have worked to improve access to college among undocumented youth. Importantly, the UndocuScholars Project is a product of a larger collaborative effort led by Suárez-Orozco and Teranishi (see Teranishi et al. 2015).

Anxiety measure
The seven-item generalized anxiety scale (GAD-7) (Spitzer et al. 2006) was used to assess clinical levels of generalized anxiety. The scale is valid for Latinx individuals in the United States (Mills et al. 2014). Participants responded to this prompt: "Over the last two weeks, how often have you been bothered by the following problems?" with sample response items including, "Trouble relaxing" or "Not being able to stop or control worrying," among others. Answer choices were rated on a four-point Likert-style scale (zero corresponded with not at all and three with nearly every day). We summed the raw scores on the seven items. This sum ranged from 0 to 21. However, one tricky aspect of this variable is that, because of an error in the survey, a small subset of respondents were not prompted to answer one of the items in the anxiety scale. Thus, we created a mean anxiety score (sum of one's anxiety score divided by the number of items answered).

Risk factors and covariates
Family responsibilities. Respondents were asked, "In a typical month, which of the following kinds of help do you PROVIDE to your family members (i.e., parents, siblings, grandparents, aunts/uncles, cousins)? (select all that apply)." There were six drop-down items: helping pay family's bills or expenses, helping family with errands/ household chores (child or elder health), tutoring or helping family members with homework or classes, translating, giving advice, or other. Each of these six items was made into a dichotomous variable, in which one meant that the student engaged in the activity and zero meant they did not. We summed each of these binary variables to create a continuous measure of the number of family responsibilities students reported.

Family motivation
This variable was based on the prompt, "My family responsibilities motivate me to continue with my college studies." Students were asked to indicate the degree to which they agreed. They could answer one (strongly disagree), two (disagree), three (neither agree nor disagree), four (agree), or five (strongly agree). If students answered four or five, we coded family motivation as one. Otherwise, it was coded as zero.

Family support
The family support variable was measured by the question, "In a typical month, which of the following kinds of help do you receive from your family members (i.e., parents, siblings, grandparents, aunt/uncles, cousins, select all that apply)?" Respondents were presented a checklist with these answers: (1) paying my expenses (e.g., housing, health or care insurance, credit card, phone bills, etc.), (2) paying for tuition, (3) helping with errands or practical tasks (i.e., rides to school or childcare, if applicable), (4) tutoring or helping me with homework and classes, and (5) helping me to solve problems/give advice. We created a sum based on whether individuals said yes (coded as one) or no (coded as zero) on each of these five items. Then, we created a categorical variable describing these groups: students who reported no support from family, students who reported receiving one or two of the supports on the list, and students who reported receiving three or more types of help from family.
Deportation worries. Students were asked, "How often are you worried that family members or friends might be detained or deported?" Answers were on a Likert scale ranging from one to four. Number one corresponded to "never" and number four to "most of the time." If students answered some or most of the time for the item mentioned above, they have a value of one for the deportation worries variable. If they did not, the value was zero.

Covariates
Other variables include a binary measure of gender, a continuous measure of age, a dummy variable indicating previous work experience, and a categorical variable of relative household income. Work experience was based on a survey question that asked respondents, "Have you had paid work experience thus far?" Available responses were yes or no, coded as a binary variable (1, 0, respectively). Relative household income was measured by a categorical variable of household income quartile representing whether respondents belonged above the 25th, 50th (median), or 75th percentile of household income. Parental education is included in the descriptive statistics to show class background. This item was based on a survey question that asked respondents about the highest level of education achieved by their mother and their father. Based on this, we created a binary measure indicating whether at least one parent had attended college.

Analysis plan
We use multivariate ordinary least squares (OLS) regression to analyze the association between DACA status and self-reported anxiety score. We add our independent variables of interest iteratively in order to examine any changes in coefficients once explanatory variables are added to the model. In addition to our successive regression models (Models 1-5 in Table 2), we add interaction terms. Interaction terms test for the joint effect of two independent variables on an outcome variable. We pursue two interaction terms because of our interest in the intersections of social locations. The first interaction term is between gender and DACA status, shown as "DACA X Men" in Table 3. This term indicates whether the association between DACA and mental health differs by gender. The second interaction term of interest is "DACA X Family deportation worries." This interaction term indicates whether family deportation worries influence anxiety differences between DACAmented and non-DACAmented students. Table 1 provides summary statistics of the key variables in the Latinx analytic sample (n = 660). Table 1 shows that over half of the analytic sample identify as women. The average age is twenty-one years old (SD: 2.65). The oldest respondent is twenty-nine years old. Twenty-seven percent of respondents report having at least one parent who is college educated. In terms of family demands and support, individuals reported having on average three (SD: 1.33) different types of family responsibilities. In terms of receiving support from family members, 7% reported not receiving any of the listed family supports (e.g., paying for expenses, paying for tuition, helping with errands/practical tasks, tutoring/helping with homework and classes, helping solve problems/give advice). Seventy-one percent reported receiving one or two types of support from their families. Twenty-two percent reported receiving three or more types of familial support. In terms of family motivation, 82% of Latinx undocumented undergraduates reported being highly motivated by family. Importantly, but not shown in Table 1, a majority Table 1 Summary of key variables, UndocuScholars Project data set 2014, Latinx undocumented undergraduates Income quartiles had relatively high missingness (3%) in the Latinx subsample. We did a simple imputation of this variable for the regression models that follow, and the tabulations in this table show the original non-imputed percentages. Income quartiles are as follows: individuals are in the top quartile if their household income was between $40,000 and 150,000; they are in the third quartile if their income is between $30,000 and 39,000; they are in the second income quartile if incomes are between $20,000 and 29,000); and they are in the first and lowest income quartile in households with incomes < $20,000. The mean anxiety score is the measure used in subsequent models of the sample are Mexican, South American, and/or Central American. In addition, over half of respondents live in California and are eligible for the California Dream Act. Seventy percent of the analytic sample are DACA recipients, and 58% reported worrying about their family's deportation. Over half reported having ever worked. Although a substantial number of students are from low-income households, there is variation. To capture the heterogeneity of income within Latinx undocumented undergraduates, we use a relative measure of income: income quartiles. The composition of students' household incomes in terms of quartiles are as follows: 18% are in the fourth quartile (household income between $40,000 and 150,000), 20% are in the third quartile (household income between $30,000-39,000), 28% are in the second income quartile (household income between $20,000 and 29,000), and 34% are in first and lowest income quartile (household incomes below $20,000). Table 2 shows the regression models that use anxiety score as the outcome variable with different predictor variables added iteratively. Model 1 includes demographic characteristics including gender, age, and household size. Model 2 adds family protective and risk factors (family responsibilities, family motivation, and family support). Model 3 adds one of the key immigration variables, DACA status, and Model 4 adds family deportation worries. Model 5 adds the socioeconomic variables ever worked and household income quartile.

Factors that predict anxiety levels
Model 1 in Table 2 shows that, consistent with previous research, students who identified as men reported lower anxiety levels compared to women. Age was associated with lower anxiety score, but this coefficient is small and not statistically significant. Household size had a positive association with anxiety score.
Model 2 in Table 2 adds family variables. Having a high number of family responsibilities does not seem to be associated with anxiety score at a statistically significant level, though students in the second quartile of family responsibilities have lower anxiety scores compared with students in the first quartile. Family motivation shows a strong negative association with anxiety score. This coefficient indicates that, compared with students who are not as motivated by their families, those who are have an anxiety score that is 0.151 points lower (equivalent to a change of one-fifth of a standard deviation in anxiety score). We do not find a statistically significant association between receiving different levels of support from family and anxiety score. We were surprised by this finding.
Models 3 and 4 include important immigration variables. Model 3 shows that compared with individuals without DACA, DACAmented individuals have anxiety scores that are 0.198 points higher, equivalent to about one-fourth of a standard deviation increase in anxiety score. Model 4 includes family deportation worries. The coefficient for this variable is statistically significant. Worrying frequently about family deportation is associated with an increase in anxiety score equivalent to 30% of a standard deviation increase in anxiety score. Notably, adding family deportation Table 2 Regression coefficients with anxiety score as the outcome, UndocuScholars Project data set 2014, Latinx undocumented undergraduates Coefficients are shown and standard errors are in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1 worries slightly attenuates the DACA coefficient, which reduces from 0.198 to 0.154. Despite attenuation, the DACA coefficient remains statistically significant. This indicates that the heighted anxiety exhibited by DACA recipients when compared with non-DACA recipients is partly (but not completely) explained by family deportation worries. An important feature of DACA is that it provided recipients a work permit and a social security number, increasing access to previously unavailable jobs and internships. It is possible that one source of stress among DACAmented students is pressure to work. When we added the work experience variable (in Model 5), we find that having ever worked is associated with higher anxiety. Adding the coefficient of the work experience variable further attenuates the DACA coefficient (reducing it to 0.0767 and rendering it insignificant at convention statistical levels). This means that DACA recipients' heightened anxiety relative to undocumented Latinx undergraduates without DACA was due to socioeconomic factors (work demands and family income) and family deportation worries.
The last variable we consider in the main models is relative income, measured by income quartile. Relative income may shape students' experiences as undergraduates because financial resources may cause stress and may limit the time students have to focus solely on school. Model 5 shows that, relative to students with families in the first/lowest income quartile, students from households in the highest income quartile had lower anxiety scores. The size of this coefficient is notable, as its coefficient is larger than the coefficient of DACA status in Model 3.
In sum, the regression models in Table 2 show tremendous complexities in the mental health of Latinx undocumented undergraduates. The main factors that seem to matter across the board are gender, family motivation, and family deportation worries. These factors retain their substantively significant associations with anxiety score even when controlling for family responsibilities, work experience, and income.
The DACA coefficient tells a complex and understudied story. Although DACA may have relieved students of some aspects of the undocumented experience such as the inability to legally work, DACAmented Latinx undergraduates have elevated anxiety levels (as seen in Models 3 and 4). We found that while DACA recipients may have relatively high levels of anxiety compared with individuals without DACA, accounting for family deportation worries and socioeconomic factors fully explains this trend (Models 4 and 5).

The role of gender
Given the extensive literature about gender differences in the expression of mental health outcomes, we examine whether the relationship between DACA and mental health varies by gender identity. We do this by adding an interaction term, "DACA X Men," to our previous model with demographic, family, immigration, and socioeconomic status control variables (Model 5 in Table 2). Results are shown in the first column (Model 1) in Table 3. For brevity, we show only the main independent variables of interest. Model 1 in Table 3 shows that the interaction term "DACA X Men" is negative and statistically significant. The negative sign shows that, for men, the relationship between DACA and self-reported anxiety is weaker compared with that of women. In other words, among DACA recipients, women have higher anxiety levels than men, but among students without DACA, levels of anxiety are more even across gender identity. Figure 1 shows the interaction term visually to ease interpretation.
The y-axis of Fig. 1 shows the linear prediction of anxiety score based on the regression in Model 1 of Table 3. The x-axis shows gender identity, and the first two bars show the predicted anxiety scores of undocumented Latinx students without DACA. The bars on the right of Fig. 1 show the predicted anxiety scores of DACA recipients. In sum, Fig. 1 suggests that women with DACA seem to have elevated levels of anxiety even when controlling for the age, household size, family responsibilities, family support, family deportation worries, work experience, and household income.
Indeed, women with DACA have higher anxiety levels compared with women without DACA and compared with men with DACA. One possible explanation for this trend is that women with DACA may face unique family pressures and family responsibilities, and may provide more emotional labor to their families because of their ability to legally work. However, even when we control for family responsibilities, this interaction persists. In light of these findings, it is important to caution interpretation of these results by stating that the outcome variable may reflect expressions of mental health. Men with and without DACA may not be as open to expressing their mental health in online surveys, which is what this study uses for the analysis. Table 3 OLS regression predicting mean self-reported anxiety score (same variables as Model 5 in Table 2) with added interaction terms Standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0. Both models include controls in Model 5 in Table 2 Variable

The role of worrying for family
The second interaction term of interest is "DACA X Family deportation worries." The coefficients of interest in this model are included in Model 2 in Table 3. The coefficients in this model are the same as those in Model 1 in Table 3, with the exception of the interaction term. Among those with DACA, those who worry about their family's deportation frequently report high levels of anxiety. The interaction coefficient is positive and is nearing statistical significance. For illustrative purposes, we also graph this interaction in Fig. 2. The y-axis is the same as that in Fig. 1, representing the linear prediction of anxiety score. The most important takeaway from Fig. 2 is that worries about family deportation do not seem to be associated with the anxiety levels of Latinx undocumented undergraduates without DACA. On the other hand, family deportation worries seem to elevate the anxiety of DACA recipients. This is important because it indicates that DACA recipients with high levels of worries about family deportation may have concerns that their families are at risk of surveillance from the state and of potential deportation. We think this is related to having DACA because non-DACAmented students with high levels of family deportation worries showed less anxiety compared with their DACAmented counterparts who were also highly worried about their family's deportation. This finding shows that, despite the promise of DACA, family-level deportation risk remains a threat to mental health. DACA in itself does not provide the tangible resources to thrive and be well.
There is a possibility that gender and worrying patterns are related. We looked more into descriptive patterns on this. Figure 3 shows that both women and men worry about their families. It does not seem to be the case that worries about family are exclusively present among women with DACA. Both men and women with  Table 3. All continuous covariates are held at their mean. The No DACA label corresponds with undocumented students not protected by the DACA program at the time of the survey (in 2014). Lines show 95% confidence intervals DACA express high levels of worry about their family's deportation. It is men without DACA that seem to report relatively low levels of worry about their family's deportation compared both to women without DACA, and to men as well as women with DACA. Relating this to our previous finding, men with DACA report slightly lower levels of anxiety, and women with DACA have significantly higher levels of anxiety. We posit that this interaction cannot be explained away by family worries. Future researchers may wish to explore this nexus further.

Heterogeneity in the undocumented experience
The contributions of this paper are multifold. First, our most unique finding is that DACA does not protect the mental health of undocumented Latinx college students. Before our analysis, we expected that DACA recipients might have reduced anxiety scores thanks to their gained access to a range of rights, a decrease in stigma, and an increase in sense of belonging. After all, their protection for deportation in 2014 (the time of the survey data collection) seemed promising. Yet, our results show that DACA recipients had heightened anxiety due to a combination of demographic factors, family deportation worries, and work experience. Women with DACA appear to have elevated anxiety scores. Why might this be, and what does it mean?
First, undocumented Latinx college students with and without DACA have undocumented family members who may be subject to all the stressors associated with their family's immigration status. Family members of Latinx undocumented college students may be exposed to extreme inequalities in accessing education as well as occupational mobility and may face excessive barriers to health care. Non-DACA recipients also have these worries, but having high levels of worries about  Table 3. All continuous covariates are held at their mean. The No DACA label corresponds with undocumented students not protected by the DACA program at the time of the survey (in 2014). Lines show 95% confidence intervals family deportation was more consequential for DACAmented Latinx college students compared with undocumented Latinx undergraduates without DACA.
The collective experience of undocumented family members and DACA recipients may increase exposure to surveillance, as DACA recipients provide the government their information such as address and history of addresses, potentially placing their family members without DACA protections at risk. In addition, DACA has always been a temporary measure (Gonzales and Ruszczyk 2021). DACA recipients may experience added stress about DACA renewals and about the future prospects of this program. Such worries, as can be seen by the recent threats to end the DACA program, would be well placed. This collective risk and worry dampens mental health (Gurrola and Ayón 2018).
It is possible that DACA recipients encounter unique stressors that add up and cumulatively make for a more stressful college experience. For instance, they may face pressures to work during college (Hsin and Ortega 2018). As our results show, having work experience helped explain the relationship between DACA and anxiety score. Importantly, DACAmented students in general may encounter acute anxiety-producing moments in their colleges because they may have to explain their liminal status as DACA recipients. For instance, one study in Colorado found that DACAmented students encountered issues during hiring processes because university personnel did not know how to handle the work permit documentation that accompanies DACA (Muñoz 2013). Moreover, all students in this study, regardless of DACA, reported rather high levels of anxiety, reflecting the overall possibility that colleges may not be prepared to provide undocumented undergraduates the resources to thrive and be well. Despite the passage of policies that facilitate access to college for undocumented students, professors may not understand the undocumented experience, campuses may have police on campus, and undocumented students may have the added stress of educating others about their undocumented experience (Muñoz 2013).
This study finds important heterogeneity in the undocumented student experience. As previously found, undocumented students differ in how they navigate their undocumented experience (Patler 2018). Factors that shape how undocumented students navigate their college experience include age of arrival and co-ethnic networks, among other factors (Patler 2018). We add to this literature by highlighting previously understudied factors that also matter. These include gender, socioeconomic status, and previous work experience. One previous study found that students with DACA may face the decision to choose work or attend school if they attend four-year universities, but that community college students with DACA adjust the units they take to accommodate their work schedules (Hsin and Ortega 2018). Interestingly, one study found that DACA recipients framed their receipt of DACA as a way to help support their families, which they reported motivated them to continue having hope that in the future they could repay their families (Luna and Montoya 2019). We suspect that the combined pressure to work during college may add stressors to the lives of Latinx undocumented college students who may already be facing institutional contexts not conducive to seeing them thrive. Given previous work that shows family contributions and responsibilities are gendered, women with DACA may have high levels of emotional and financial responsibilities to their families. This may be one reason for their elevated anxiety scores. The findings on anxiety scores of Latinx DACAmented women underline the need to support college student mental health using a gender-sensitive approach (Ai et al. 2015).
This study contains limitations. First, our data may not be representative of the Latinx undocumented college student population at the national level. It is possible that students with high levels of anxiety about their immigration status may be less likely to partake in surveys in general. If this is the case, our findings would underestimate anxiety scores. It is important to note that this study is not an assessment of clinical diagnoses. In addition, we did not capture potentially important differences in campus cultures and institutional resources by campus type. Recent research in higher education indicates these factors matter for student outcomes and resources that may shape student well-being (Garcia 2019;Reyes 2018). Last, we did not capture the mental health and gender relationship beyond the gender binary, and we encourage future researchers to include expansive gender identities thoughtfully in their surveys of undocumented individuals, as more recent surveys have done (Enriquez et al. 2020).
Despite limitations, this study holds important implications for data collection on Latinx college students. Research about Latinx college students that does not include information about family's deportation risk may miss important family-level factors that shape student well-being. One in twenty US-born children live in mixedstatus families (Passel et al. 2018). Thus, the present study on undocumented Latinx college students demonstrates the possibility that US-born Latinx students' mental well-being may be threatened if they have undocumented family members.
Our research suggests that DACA alone does not provide young adults with tangible resources that protect against anxiety, such as authorization to reside in the United States, and it does not protect them from worrying of others' risk of deportation. These findings help immigrant advocates argue that it is not enough to maintain the DACA program; undocumented immigrants are facing significant stress and angst, which must be relieved through other pathways of citizenship and/or liberation for immigrants, as well interventions to halt deportations.
The DACA program has been a target of the Trump administration. Although the data used in this study was collected in 2014, it may reflect a conservative estimate of the mental health of undocumented Latinx college students. Arguably, the political climate and the uncertainty about the future of DACA may heighten the anxiety of undocumented Latinx college students. In addition, the overall anti-immigrant presidential administration from 2016 to 2020 may have caused extreme stress to the broader Latinx as well as immigrant community with family ties to undocumented individuals.
We also find some promising results. Students who reported being motivated by their family have more positive mental health outcomes. This suggests that family may be a positive source of meaning-making and may represent a form of community cultural wealth (Yosso 2005) for undocumented Latinx college students. It also shows that, in response to volatile political contexts, Latinx college students forge ways of coping and resisting (Castrellón et al. 2017). In sum, the mental health of undocumented Latinx students is complex. There is heterogeneity emerging from multiple dimensions including gender, family motivation, DACA status, family deportation worries, and socioeconomic status. Given the current political moment and recent uncertainty with DACA, the wider public may need to be reminded that the promise of DACA may have been short-lived, and that permanent paths to legal status for undocumented young adults and their families may be necessary to promote their mental health, which is important for the retention of Latinx undergraduates in higher education.