Debt Stress, College Stress: Implications for Black and Latinx Students’ Mental Health

Abstract

Educational debt is an economic stressor that is harmful to mental health and disproportionately experienced by African American and Latinx youth. In this paper, we use a daily diary design to explore the link between mental health, context specific factors like “college stress” and time use, and educational debt stress, or stress incurred from thinking about educational debt and college affordability. This paper utilizes data from a sample of predominately African American and Latinx college students who provided over 1,000 unique time observations. Results show that debt-induced stress is predictive of greater self-reported hostility, guilt, sadness, fatigue, and general negative emotion. Moreover, the relationship may be partly mediated by “college stress” reflecting course loads and post-graduation job expectations. For enrolled students then, educational debt may influence mental health directly through concerns over affordability, or indirectly by shaping facets of college life. The window that our granular data provides into college experiences suggest that the consequences of student debt are manifest and immediate. Further, the documented day-to-day mental health burden for minority students may contribute to downstream processes like matriculation.

Introduction

Cumulative student debt is now the most substantial type of non-mortgage debt in the United States (Friedman 2020). Arguably the end result of skyrocketing costs, state defunding of higher education institutions, and stagnant familial resources (Bound, Lovenheim, and Turner 2007; Houle and Addo 2018), college students are increasingly using debt to meet the financial costs of college. Estimates suggest that the average debtor leaves school with more than $30,000 in student debt (Reed and Cochrane 2012; Student Borrower Protection Center 2020). Given these trends in rising education costs, researchers have recently begun examining the social and health consequences for indebted students, including how debt is associated with engagement in college activities (Quadlin and Rudel 2015), and the relationship between debt and mental health among university undergraduates (Cooke et al. 2004; Walsemann, Gee, and Gentile 2015). A key emphasis of this research is that student loan debt is a stratifying mechanism for students’ time use, the foundation of campus life, and their mental health.

When assessed by race, data suggests that Black and Latinx students may experience even more debt stress than their White counterparts. For example, Blacks on average owe $5,000 to $10,000 more than White debtors from otherwise comparable backgrounds (Houle and Addo 2018; Huelsman 2015). While there is very little data about borrowing for Latinx groups in higher education, one study shows higher borrowing rates in certain university contexts when compared to Whites (Huelsman 2015). What remains underexplored however, is how heavy debt burden and context specific factors, like college stress and time use, shape the mental health of minority students on a day-to-day basis.

In this paper, we utilize a two-week daily diary design to link fluctuations in perceptions of educational debt to mental health in a sample of Black and Latinx college students while debt is being accumulated. We treat educational debt as an economic stressor, hereafter referred to as educational debt stress, and define it as stress incurred from thinking about educational debt and college affordability. We posit that it may directly influence mental health through concerns about affordability or indirectly by qualitatively shaping the college experience. Although the realities of educational debt and college stress are not specific to racial minorities, these factors may contribute to stress that decreases the quality of educational experiences and could ultimately contribute to differential matriculation rates (Cook & Cordova 2006; Kim 2007; Chen and Desjardins 2010).

Building on stress-related mental health research and the nascent literature surrounding race and student debt, we therefore sought to test four hypotheses: First, that educational debt stress is associated with decreased mental health among Black and Latinx youth. Second, that this association is mediated by “college stress” reflecting course loads and post-graduation job expectations. Third, that educational debt- and college-stress associations with mental health arise from variation in students’ study and work life schedules. Finally, because prior work suggests that family income may moderate the debt stress association, we also assess the extent to which household income moderates mental health associations with educational debt and college stress.

To our knowledge, this study is novel in that our analysis emphasizes the links between debt stress, “college stress”, and mental health in a predominately minority sample. Moreover, we contribute to a nascent literature on “predatory inclusion” in which minority students are provided access to higher education under conditions that jeopardize the benefits of access (Seamster and Charron-Chénier 2017). While in its original conception the theory focuses on long-term financial returns, or a lack thereof, we argue that predatory inclusion may be extended to consider the immediate health costs incurred by minority students as they navigate college accruing debt.

Literature Review

American society is a “racialized social system” (Bonilla-Silva 1997) where systemic racial inequalities span nearly all social domains (Reskin 2012). It is likely unsurprising then that within the context of higher education, racial discrimination shapes who feels debt as a crushing burden and who experiences debt as an opportunity (Seamster 2019). Consequently, racial and ethnic minorities are at a disproportionate risk of bearing the burdens of high debt and subsequent outcomes (e.g., loan default, decreased capacity to plan financially). Accumulating such an extreme financial burden, as they navigate the daily stressors and idiosyncrasies of college life, is likely to influence how minority students cognize and experience educational debt and, importantly, how these processes affect mental health.

Stress Process Model

To understand how educational debt may be particularly consequential for minority mental health, we orient this study around the Stress Process Model (SPM), which combines three major conceptual domains: sources of stress, mediators of stress, and manifestations of stress (Pearlin et al. 1981). Within this framework, individuals are first exposed to a stressor that arises from the occurrence of a distinct event, or the presence of an on-going problem (chronic strain). The increasingly complex and connected lives that people lead however, suggests that stressors, whether in the form of events or durable strains, do not exist in isolation from other problems but instead operate as a package or constellation (Pearlin 1989). To capture this full array of stressors contemporaneously present in individuals’ lives, researchers distinguish between primary and secondary stressors (Pearlin 1989), as well as direct and indirect effects of stress (Pearlin and Bierman 2013). A critical innovation of this distinction is the ability “to discriminate between people who may be similar with regard to their exposure to one stressor but who differ appreciably with regard to the array of stressors to which they are exposed” (Pearlin 1989:248).

To contextualize how and why some people seem to differ markedly in the intensity and range of stress outcomes they manifest, the SPM considers how the physical and mental health consequences of stress are shaped by individuals’ advantaged or disadvantaged positions in the social structure (Thoits 2006). The model posits that one’s placement within a status hierarchy regulates stress exposure and therefore psychosocial health risks (Glass and McAtte 2006; Pearlin 1999). Adopting the SPM as a guiding framework provides an opportunity for this study to explore how educational debt, an economic stressor, impacts the mental health of minority students, the disproportionate debt bearers. Further, the stress process paradigm acknowledges that stressors bundle together and exert influence on one another in daily life, allowing us to simultaneously consider how “college stress” and time use work in tandem with educational debt stress to shape the mental health profiles of minority students.

The Relationship Between Educational Debt and Health

A growing body of research documents a strong and persistent link between financial strain and mental health (Kahn and Pearlin 2006; Drentea and Reynolds 2012), including personal indebtedness and psychological functioning (Selenko and Batinic 2011), anxiety (Cooke et al. 2004), mental disorders (Jenkins et al. 2008), and depressive symptoms in later life (Aranda and Lincoln 2011).

Utilizing large-scale quantitative surveys of students to assess whether a correlation exists between student debt and a range of health indicators (for review see Nissen et al. 2019), studies of educational debt on mental wellbeing typically indicate that students with higher levels of financial concern tend to have worse mental health scores (Richardson et al. 2017), assessed via standardized measures like the Generalized Anxiety Disorder questionnaire (GAD-7; Spitzer et al. 2006) or Perceived Stress Scale (PSS; Cohen et al. 1983), and poorer psychological functioning (Walsemann, Gee, and Gentile 2015). In evaluating this relationship, a consistent finding across many studies is that perceptions of debt may be more consequential for mental health than actual levels of debt (Nissen et al. 2019).

Stradling (2001), for example, found that students who perceived their anticipated graduate debt as ‘excessive’ were more likely to be anxious or depressed than students who viewed their anticipated debt as ‘manageable’. Furthermore, among college-enrolled students, those classified as “high debt worry” report feeling more “tense, anxious, and nervous” (Cooke et al. 2004) and have more difficulty getting sleep (also Walsemann, Gee, and Gentile 2015; Richardson et al. 2017). Beyond negative impacts to borrower’s health, student perceptions and loan borrowing experiences have been associated with credit card debt, decreased capacity to plan financially, and extended years to homeownership (Allen et al. 2020; Cho et al. 2015; Fox et al. 2017; Henager and Wilmarth 2018). In a study of debt perceptions among Black students, Baker (2019b) found that students with higher levels of debt reconsidered their postsecondary and career plans. Related, Malcom and Dowd (2012) show that, irrespective of race/ethnicity, perceptions of college debt may discourage students from continuing their education in a graduate program. Taken together, these findings suggest that perspective toward student debt may be a key pathway for understanding how debt also affects health.

Yet, worry and stress associated with educational debt are likely to be experienced differently across racial and economic lines. This variance may be attributable to racial differences in family socioeconomic status (Jackson and Reynolds 2013; Ratcliffe and McKernan 2014), reliance on student loan debt (Cunningham and Santiago 2008), and risk of loan default (Gross et al. 2009). Research on Latinx experiences with student loans emphasizes their aversion to borrowing and notes that a lack of familial capital plays a major role in Latinx decisions to borrow (Boatman et al. 2017; Cunningham and Santiago 2008). Accordingly, the current paper focuses on data from a minority population because they are disproportionately burdened with debt (Addo et al. 2016; Houle and Addo 2018), are more likely to come from economically disadvantaged families, and thus are at heightened exposure to the risks of indebtedness (i.e., high debt burden, credit consequences, health costs). In fact, in a recent study of debt stress and mental health among an ethnically diverse sample, scholars more consistently found significant, negative associations for Black and Latinx students compared with Asian and, to a lesser extent, white students (Tran et al. 2018). As research on the relationship between educational debt and health in Latinx populations remains limited, this study provides an opportunity to glean information on an understudied group. Noting that the stressor used in this study is not objective debt, but debt stress (stress associated with thinking about college debt and affordability), we hypothesize that educational debt-induced stress will be positively associated with a range of negative affective states among a sample of predominately Black and Latinx students.

“College Stress” and Time use as Potential Links Between Debt and Mental Health

In addition to educational debt, college students encounter a unique set of experiences that may 1) evoke stress and 2) are particularly consequential for indebted students. We refer to this facet of college life as “college stress” and operationalize the term using course load, perceptions about course rigor and magnitude of homework, and post-graduation job expectations. Additionally, time use allows us to tap into student involvement, within and across types of life activities, to contextualize the college experience and the multitude of factors potentially shaping the relationship between debt and mental health.

Course Load

Academic pressures have been regarded as the most common stressor for college students (Ackgun and Ciarrochi 2003; Barker et al. 2018; Murff 2005; Yang et al. 2021), and on any given day may take the form of learning and examination, performance competition, requirement to master much knowledge in a short period of time, and rigorous course loads, or the number and difficulty of courses taken in an academic setting (Montgomery and Côté 2003; Mishra 2017). This study focuses on course load because it has been linked to health among college students (Conley and Lehman 2011; Papousek et al. 2010) and is an indirect, yet repeated way in which students encounter a facet of educational debt. Given that courses are a primary expense covered by student loans, it is probable that the amount of stress students feel about debt is influenced by how they perceive and experience courses. Accordingly, our definition and operationalization of course load reflects student perceptions of course rigor and magnitude of homework. If students are amassing large amounts of debt for courses perceived to be stressful, mentally taxing, and time-consuming, it is likely to adversely shape their mental state.

Post-Graduation Job Expectation

College students' sense of control may seem precarious given uncertainty in the utility of their degree, their ability to pay off student debt, and their future job prospects and economic earnings. Research suggests that for socially and economically disadvantaged students such feelings are exacerbated (Bean and Eaton 2000). This may be especially true within Black and Latinx student populations. For example, higher unemployment rates are faced by blacks relative to whites at every level of education (Jones and Schmitt 2014). Among college graduates, blacks have lower earnings (Zhang 2008), more student debt (Huelsman 2015; Houle and Addo 2018) and report greater concern about the affordability of student loan payments (Ratcliffe and McKernan 2013). Latinx households dedicate a larger percentage of their familial income to college (Young Invincibles 2017), while Latinx graduates are faced with the lowest post-graduation earnings of all racial/ethnic groups (Unidos 2019b). Consequently, accumulating debt within a society of racialized and uneven degree returns is likely to impact the amount of stress incurred from such debt, and ultimately mental health. Therefore, we hypothesize that the negative mental health influence of educational debt stress is partly explained by “college stress”.

Time Use

Despite the finite number of hours in a day or week, students use their time in vastly different ways, opting into and out of a range of academic, work-related, social, and extra-curricular activities. Two perspectives on debt and time use have emerged out of the larger tradition of research on college, debt, and emerging adulthood (Quadlin and Rudel 2015). The first suggests that indebted students may acquire a greater sense of responsibility, spending more time on academics, work, or other activities that could increase their success on the job market post-graduation (Bodvarsson and Walker 2004; Dwyer et al. 2011). A second perspective positions debt not as responsibility, but as a liability, and finds that the financial constraints associated with debt may exclude students from some aspects of college life (Drentea 2000; Espenshade and Radford 2009). Quadlin and Rudel (2015), finding a mix of both perspectives, position debt as a polarizing experience – pushing some students to dedicate most of their time to working for pay, going to class, and studying, while encouraging others to withdraw from campus life altogether (Quadlin and Rudel 2015). We extend this work by exploring how time use, a window into the college experience, may provide nuance to the relationship between debt and mental health.

A second way that we extend this work is by exploiting the temporal nature of our data to examine time use in day-to-day life. Time use measures are well suited for studying how individuals vary in everyday experiences and have been used extensively by scholars to explore gendered differences in household labor (see Milkie et al. 2009; Sayer et al. 2004; Schneider 2012), for example. Yet, as noted by Quadlin and Rudel (2015), variation in time use among college students remains relatively underexplored. A handful of surveys, conducted at one time point, retrospectively inquire about time spent in typical college activities (Arum and Roksa 2011; Brint and Cantwell 2010; Stinebrickner and Stinebricker 2004). Additional studies increase their accounts of time use by utilizing longitudinal surveys in which students are asked annually about time use (i.e., National Longitudinal Survey of Freshman; see Massey et al. 2006; Quadlin and Rudel 2015) or through analyzing multiple datasets from different time periods (Babcock and Marks 2011). Still, these longitudinal studies are limited in that the time use measures are retrospective and subject to recall bias and other inaccuracies (Quadlin and Rudel 2015). We add to this body of work by offering time use data from a 2-week daily diary study reported by predominantly minority students. In addition to reducing error through data collection closer in time to actual experience, we use detailed time use diaries to provide context to the minority student college experience. Doing so may provide nuanced understandings of the debt-mental health relationship.

Role of Income

A handful of studies have highlighted the influence of social class or parental resources on student attitudes towards financial affairs and debt. While a UK based study suggests that parental wealth may act as a buffer for financially well-off students (Callender 2003; Callender and Mason 2017), work by Walsemann and colleagues (2015) in the U.S. context finds that among wealthier families the cumulative amount of student loans borrowed was associated with poorer psychological functioning. Among less wealthy families the reverse association was found, such that increasing student loans was related to better psychological functioning (Walsemann, Gee, and Gentile 2015). Research that considers both income and race finds that across all income quartiles, Hispanic and Asian students were less likely to borrow than White students, while Black students were the most likely to borrow across income brackets and institutional types (i.e community, public, private, for-profit; Cunningham and Santiago 2008). Based on these studies, we assess the extent to which income moderates the associations between debt stress and “college stress” and mental health. However, given the differing results and our utilization of household income as opposed to wealth, we do not offer specific directional hypotheses.

The Present Study

In this study, we situate educational debt as a stressor that is commonly experienced by minority students, and one that creates risks for minority mental health. We also consider educational debt not as an isolated stressor, but one that is bundled with “college stress” and time use to conjointly shape the mental health of currently enrolled students. Stressors are dynamic and they ebb and flow in frequency and intensity. Accordingly, a novelty of our study is the exploitation of high intensity longitudinal data over short periods of time so that the fluctuations of multiple stressors (educational debt and college stress) and time use can be linked to fluctuations in mental health.

Leveraging this temporality, we quantitatively peer into the qualitative college experience to explore the day-to-day mental health burden of accumulating debt while navigating curriculum, busy schedules, and post-graduation expectations. To date, few studies have explored how omnipresent educational debt is in students’ consciousness. This study moves us closer to that objective by exploring just how stressful debt is—directly through affordability and indirectly through college experience.

Method

Data

Data for this study come from the StudentHD2 project, which was conducted on a large Midwestern research university campus during the Fall of 2017 and Spring of 2018. The university is a predominantly white institution, with an undergraduate student population comprised of approximately 72% white 7% Latinx, 3% Black, and 3% Asian students (n = 25,327). The goal of StudentHD2 was to examine the dynamic experiences of psychological, physiological, and behavioral outcomes associated with a wide range of stressors among predominately racial/ethnic minority students. Students participated in intake and exit interviews sandwiching a 2-week daily diary protocol with a detailed morning diary administered through smart phone that interrogated student experiences and feelings over the previous day. Having students document their responses the following day—as opposed to the same evening- increased the range of experiences and responses that could be captured by facilitating inclusion of events happening at night. This study draws on data collected from the intake and detailed daily diary surveys. The full sample was comprised of 83 Black, Latinx, and “other” students who collectively contributed 1011 unique time observations via the surveys. The daily diary response rate was 87% with students providing 12.8 days on average. All study procedures were approved by the University IRB.

Measures

Dependent variables

The mental health measures are from the PANAS-X (Positive and Negative Affect Schedule), a 60-item version of the PANAS that includes thirteen affective states for mood measurement. While the core order dimensions of Positive Affect (PA) and Negative Affect (NA) reflect global valence, the sub-scales reflect specific affective states. For this study, five scales were utilized that include one of the broader, higher order dimensions, general negative emotion (10 items, α = .491), and four of the lower-level scales: hostility (6 items, α =.73), guilt (6 items, α = .981), sadness (5 items, α = .800), and fatigue (4 items, α = .827).¹ Psychometric properties of PANAS scores have been found valid and reliable in community samples of African Americans (Merz et al. 2013) and multiethnic samples including Latinx individuals (Villodas et al. 2011). In this study, students were asked to what extent (none = 1, little = 2 some = 3, and lots = 4) they felt each of the items (emotions) on the previous day. Responses for each item were then averaged and standardized to create five standardized emotion scales. To ensure that items within scales were equally weighted, each summed total was divided by the number of nonmissing items present. If less than half of the items were present, a scale was coded to “missing” for that day.

Key variables

The main independent variable used in this analysis is educational debt stress, or stress incurred from thinking about educational debt and college affordability. While debt is arguably a constant stressor for the duration of the study period, students report considerable variation in stress felt from thinking about debt. Drawn from the Inceptia Financial Stress survey (2012), the four items used to construct this variable reflect the top stressors reported by a nationally representative group of college students and recent college graduates (Inceptia 2012). In this study, students were asked daily “how stressed out they felt thinking about” 1) having enough money for school, 2) borrowing money for school, 3) the need to repay loans, and 4) the cost of education. We conducted a factor analysis and found that all four measures substantially loaded onto one factor (all loadings > .79, α = .959). Thus, the mean scale used in the analysis reflects the student’s averaged daily responses to the four items. Daily responses were captured in a truncated scale (none = 1, some = 2, and lots = 3) to facilitate viewer readability and accessibility on their smartphones. The scale was standardized ($\overline{x}=0$, σ = 1) for the analysis.

The “college stress” measure was constructed using two additional items from the Inceptia Financial Stress survey and a third item that all loaded substantially onto one factor (all loadings > .58, α = .794). Specifically, students were asked daily “how stressed out they felt thinking about” 1) the need to get a job after college, 2) how challenging courses are, and 3) the amount of homework they had to complete. Daily responses (none = 1, some = 2, and lots = 3) across the three items were again averaged and standardized to create a final standardized mean scale.

Time use reflects student’s daily reported hours involved in the following 4 activities: class, studying, working, and relaxing. Specifically, students were asked to identify each 30-minute interval between 6am and 12am (with the assumption that sleep was occurring in the remaining 6 hours of the 24 cycle) that they participated in all of the four activities on the previous day. Because it is possible that a student participated in multiple activities within a 30-minute interval, they were allowed to select as many options as needed. Each interval then became a dichotomous variable (1 = participation), that could be summed across the day and divided by 2 to reflect total number of hours involved in each activity. Final analysis includes up to 14 days of information on time use for each activity.

Last, literature suggests that income and wealth may act as a buffer, or moderate, the relationship between educational debt and mental health (Callender 2003; Walsemann, Gee, and Gentile 2015). To test this hypothesis, the study used income information collected during intake. Students were asked to provide their best estimate of their family household income. If less than $100,000, students were able to select the most appropriate $10,000 range interval (ex. $20-000, $29,000). Above $100,000, the increment scaling increased. Specifically, students could report household incomes between $100,000 and $149,999 or greater than $150,000. Respondents were able to select from a total of 12 intervals (as shown in the Appendix, the distribution covered the entire range of intervals). The income distribution was recoded into both a median split and tertiary split. Median split is a binary coding where respondents are coded as “1” if reported income falls below the median income value. In the tertiary split, respondents are coded as “−1” if reported income falls in the bottom third of the income distribution, “0” if reported income falls in the middle third of the income distribution, and as “1” if reported income falls in the top third of the income distribution. Interactions were then assessed using multiplicative terms with the other key variables.

Analytic Strategy

This study employs a within participants fixed effect model with robust standard errors to assess whether educational debt-induced stress is associated with mental health among college students from day-to-day. We then added “college stress”, and time use, before testing income interaction effects to assess the potential moderation. This study therefore relies on within-person variation while using fixed effects to control for all between-participant effects that were stable over the study period (e.g., this approach implicitly controls for actual debt). Final models are estimated after listwise deletion (only missing values across the dependent variables omitted). The final analytic sample sizes differed according to the mental health outcome estimated. Sample sizes are as follows and indicative of repeated moments reported by individuals: general negative emotion (n = 947), hostility (n = 898), guilt (n = 897), sadness (n = 951), and fatigue (n = 896).

The models are structured over time points t for each participant i. The outcome variable, y_ti, is standardized (${\overline{x}}_l=0$, σ_i = 1) within participants and the final main effects model (Model 3) was based on equation 1:

$$\begin{gathered} y_{ti}={\beta }_0+{\beta }_1{x}_{ti}+{\beta }_2{w}_{ti}+{\beta }_3{c}_{ti}+{\beta }_4{s}_{ti}+{\beta }_5{k}_{ti}+{\beta }_6{r}_{ti} \\ +p_i+t_t+e_{ti} \end{gathered}$$ (1)

Here β₀ is the model intercept, β₁ captures the main effect of educational debt-induced stress (x_ti) and β₂ is the main effect of “college stress” (w_ti). β₃ to β₆ then capture the main effects of time spent in class ( c_ti), studying (s_ti), working and relaxing (r_ti). Person fixed effects are included in p_i, time-fixed effects for each day of the week are represented in t_t, and the residual is denoted as e_ti. This model effectively controls for all unmeasured differences between individuals that do not change over the study participation period. Model 2 omits c_ti, s_ti, k_ti, and r_ti to represent the main effects of educational debt stress and “college stress”, and the baseline model (Model 1) omits w_ti to only represent the main effect of educational debt stress.

The income interactions were assessed using model F-tests for each outcome to assess if adding an income interaction effect improved model fit. For Model 4 (equation 2), a new term β₇x_ti * I_i the interaction effect between income and educational debt stress, was added. The main effect of income cannot be included because of the inclusion of the person fixed effects, p_i, which adjust for all between-student covariates, such as income and actual debt. This model is a difference-in-difference estimator. In other words, the difference in the effect of x_ti by a between-student covariate, I_i. Model 5 (equation 3) then added β₈w_ti * I_i to Model 4. Model 4 was compared to Model 2 and Model 5 was compared to Model 4. We make comparisons against Model 2 because Model 3 provides largely null results. Specifically, because the time use variables included in Model 3 (equation 1) did not consistently predict our mental health measures, we used the more parsimonious specification.

$$y_{ti}={\beta }_0+{\beta }_1{x}_{ti}+{\beta }_2{w}_{ti}+{\beta }_7{x}_{ti}\ast I_i+p_i+t_t+e_{ti}$$ (2)

$$y_{ti}={\beta }_0+{\beta }_1{x}_{ti}+{\beta }_2{w}_{ti}+{\beta }_7{x}_{ti}\ast I_i+{\beta }_8{w}_{ti}\ast I_i+p_i+t_t+e_{ti}$$ (3)

Results

Descriptive Statistics

Descriptive statistics for the 83 participants are presented in Table 1. The sample includes the following race/ethnic groups: 54 African American/Black, 24 Hispanic/Latinx, and 5 others (3 = other, 1 = Continental African, 1 = White). Approximately 65% of the sample identified as African American/Black, 29% as Hispanic/Latinx, and 6% were classified as other for the purpose of the study. 60% of the sample identified as female (n = 50) and the mean age of respondents was 20.01. Average years in school for participants was 14.25 years with the following class year distribution: 31 freshman, 24 sophomores, 11 juniors, 11 seniors, and with 6 students identifying as 5th years or higher.

Table 1:

Descriptive Sample Statistics (N = 83)

	Freq.	%/Mean	SD	Min	Max
African American	54	65
Hispanic/Latinx	24	29
Other	5	6
Median Income and below (40k-49k)	44
Female	50	60
Age		20.01	8.14	18	33
Year in School		14.25	1.34	13	18
Days in Study		12.18	5.12	1	15

Descriptive statistics for the daily diary portion of the study are presented in Table 2. The mental health scales are all right skewed since most students reported relatively low values on the scales overall. Mean responses ranged from 1.333 (guilt) to 2.020 (fatigue) on a scale ranging from 1-4. In terms of variability, across the 5 mental health scales, ICC (intraclass correlations) ranged from .43 (fatigue) to .56 (guilt), such that roughly half of the variation in reported emotions stem from within-person variability. Reported educational debt stress and “college stress” had averages of 1.316 and 1.591 (scale ranging from 1-3), with 33% and 48% of variation within-person respectively. For measures of time use, reported mean hours range from 2.977 (class time) to 10.89 (relax time). Compared to other measures, time use displayed a high degree of within-person variability ranging from 61% (relax time) to 94% (class time).

Table 2:

Time Variant Variables (N = 83, n_t = 1011)

	Unstandardized mean	Min	Max	Possible Range	ICC
Dependent Variables
General Negative	1.409	1	3.625	1-4	.52
Hostility	1.338	1	3.750	1-4	.48
Guilt	1.333	1	4	1-4	.56
Sadness	1.438	1	4	1-4	.46
Fatigue	2.020	1	4	1-4	.43
Independent Variables
Educational Debt Stress	1.316	1	3	1-3	.67
College Stress	1.591	1	3	1-3	.52
Time Use - Class	2.977	0	18	18	0.06
Time Use - Studying	4.010	0	18	18	0.27
Time Use - Work	3.601	0	16.5	18	0.30
Time Use - Relax	10.89	0	18	18	0.39

Notes: ICC values reflect between-person variation

Model Results

Table 3 presents the fixed effect results for general negative emotion, hostility, guilt, sadness, and fatigue. The first model within each outcome/grouping includes within-person measures of educational debt stress (stress incurred from thinking about educational debt). The next model then adds the mediating role of “college stress”. The final model in each set then adds the potential effect of time use. We first tested whether educational debt stress was associated with mental health in a sample of predominately Black and Latinx students. The results displayed in the first column within each outcome (Table 3) show that reported stress incurred from thinking about educational debt and college affordability is associated with a broad pattern of poor mental health. Specifically, a 1 standard deviation (SD) increase in educational debt stress was associated with a .191(***) SD increase in general negative emotion, a .164(***) SD increase in hostility, a .208(***) SD in guilt, a .200 (***) SD increase in sadness, and a .172(***) SD increase in fatigue. The consistent pattern of positive association across the five mental health dimensions aligns with the broader literature on health and financial strain (Kahn and Pearlin 2006). Our contribution is detection of this association among minority students in daily life. We find that while students are not constantly thinking about debt, they tend to feel worse when they do. When we re-estimated the analytic models with only Black and Latinx students (n = 78; 5 others omitted), the results were consistent with those reported here (see Appendix).

Table 3:

Person and Time Fixed Effects Models for Educational Debt Stress and “College Stress”

	General Negative			Hostility			Guilt			Sadness			Fatigue
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)
Educational Debt Stress	0.191*** (0.042)	0.130** (0.042)	0.125** (0.042)	0.164*** (0.041)	0.117** (0.043)	0.117** (0.044)	0.208*** (0.049)	0.155** (0.050)	0.153** (0.050)	0.200*** (0.044)	0.167*** (0.046)	0.167*** (0.046)	0.172*** (0.045)	0.082 (0.046)	0.085 (0.046)
College Stress		0.158*** (0.037)	0.141*** (0.038)		0.114** (0.041)	0.105* (0.041)		0.136*** (0.041)	0.130** (0.043)		0.082* (0.038)	0.079* (0.038)		0.242*** (0.042)	0.235*** (0.042)
Time Class			0.038 (0.033)			0.037 (0.036)			−0.0004 (0.029)			0.065 (0.039)			−0.017 (0.044)
Time Study			0.097 (0.042)			0.043 (0.042)			0.025 (0.041)			0.061 (0.048)			−0.045 (0.044)
Time Work			0.006 (0.039)			0.028 (0.036)			0.022 (0.037)			0.054 (0.040)			−0.007 (0.044)
Time Relax			−0.005 (0.054)			0.004 (0.055)			−0.005 (0.054)			0.093 (0.067)			−0.143* (0.064)
Constant	−0.700*** (0.073)	−0.563*** (0.079)	−0.596*** (0.106)	−0.543*** (0.079)	−0.444*** (0.083)	−0.501*** (0.108)	−0.586*** (0.092)	−0.465*** (0.091)	−0.490*** (0.111)	−0.436*** (0.086)	−0.366*** (0.094)	−0.335** (0.123)	−0.954*** (0.180)	−0.746*** (0.187)	−0.974*** (0.202)
Fixed effects?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Nt	947	947	947	898	898	898	897	897	897	951	951	951	896	896	896
Adjusted R 2	0.537	0.547	0.552	0.488	0.493	0.492	0.572	0.579	0.577	0.474	0.476	0.476	0.473	0.497	0.501

^*p< .05.

^**p< .01.

^***p< .001.

Notes: Data for this study come from the StudentHD2 project. Observations are drawn from a sample of 83 students. Values in parentheses are robust standard errors.

Next, we included “college stress” to examine whether it explains the link between debt stress and mental health. Recall that this hypothesis predicts that stress stemming from course loads and post-graduation job expectations, all key facets of the modern college experience, will influence how students cognize debt and in turn, operate as an additional pathway by which debt impacts student mental health. The results displayed in the second column within each outcome group support a role for “college stress” in understanding the consistent link between stress incurred from educational debt and mental health. In providing tentative, empirical support for both a direct (concerns about affordability) and indirect (shaping facets of college life) daily influence of debt on minority mental health, we add a novel finding to the current literature. Specially, when “college stress”, a previously unconsidered path, is added to the models the relationship between educational debt stress and mental health is attenuated. There is a reduction in the effect size of the coefficient associated with sadness (.167***), and the general negative emotion, hostility and guilt measures lose levels of significance in the presence of “college stress” (.130**, .117** and .155** respectively). Further, the relationship between educational debt induced stress and fatigue was nonsignificant (.082). It is also key to note that there is a consistently significant effect of “college stress” across all five outcomes. The association of “college stress” with mental health was similar in magnitude to the debt stress coefficients, with the exception of fatigue which showed a large effect size (.242***) and may be indicative of the physical and mental exhaustion accompanying course loads for many students.

A formal mediation analyses support our claim that “college stress” partially mediates the relationship between educational debt stress and mental health. As Table 4 illustrates, the indirect effects were statistically significant (p< .05) for all five emotional scales. That is, “college stress” explained a significant portion of the effect of educational debt stress on mental health. Standardized indirect effects were computed for each of 1,000 bootstrapped samples, and the 95% confidence interval was computed by determining the indirect effects at the 2.5^th and 97.5^th percentiles.

Table 4:

Mediation Analyses for “College Stress” in Relationship Between Educational Debt Stress and Mental Health Outcomes

	Indirect Effect				Proportion Mediated
	Estimate	P-value	CI Lower	CI Upper	Estimate	P-value	CI Lower	CI Upper
General Negative	0.0612	***	0.03	0.09	0.3179	***	0.15	0.58
Hostility	0.0458	**	0.01	0.08	0.2897	**	0.09	0.69
Guilt	0.0526	***	0.03	0.08	0.2538	***	0.11	0.47
Sadness	0.0329	*	0.004	0.06	0.1616	*	0.02	0.39
Fatigue	0.0902	***	0.06	0.13	0.5201	***	0.29	1.11

^*p< .05.

^**p < .01.

^***p < .001.

Notes: Estimates for indirect effects reflect the product of the effect of the independent variable on the mediator and the effect of the mediator on the dependent variable (captured in the even numbered columns of Table 3). Confidence intervals (95% level) and significance levels are for the entire indirect effect.

We then included measures of time use to examine whether fluctuation in time spent in class, studying, working, or relaxing also influenced mental health among students. Considering our study’s focus on within-person variation in experiences, time use could be both a large source of variation and independent stressor influencing mental health. Nevertheless, the results displayed in the final column of each outcome offer minimal support for a potential effect of fluctuating work-life schedules on mental health. Hours spent across these four domains are largely non-significantly associated with all 5 of the assessed mental health outcomes, suggesting that day-to-day variation in students’ schedules are unrelated to mental health fluctuations.

Next, we examined whether income moderated the relationship between educational debt stress and mental health, as well as the relationship between debt stress and “college stress”. The null findings around time use prevented us from exploring their interaction within income. Because income is time-invariant or stable for the duration of the study, its main effect is already captured in the person fixed effects. As such, we added interaction effects between income and educational debt stress, and income and “college stress”, to assess if the difference in the reported stress incurred thinking about debt (or the differences in the reported measures included in “college stress”) differed by income level (difference in difference). The results of this analysis, which are presented in Table 5 and 6, show that the moderating effect of income on educational debt stress and “college stress” is small and statistically insignificant. Income moderation is explored in two ways. First, with a median split where income is coded as either above or below the mean. Second, with a tertiary split where income is divided into the lowest, middle, and highest third of the income distribution. Table 5, reflecting the median split, and Table 6, featuring the tertiary split, both assess if adding the independent interaction effects of income and educational debt stress and income and “college stress” improve the overall fit for each model within each mental health measure.

Table 5:

F-test for Significance of Income (median split) Interaction

Outcome	xti * Ii		Wti * Ii
	F	P-value	F	P-value
General Negative	0.5134	0.4739	0.2533	0.6149
Hostility	0.0567	0.8119	0.7432	0.3889
Guilt	0.2706	0.6031	0.0987	0.7535
Sadness	1.8475	0.1744	0.0254	0.8733
Fatigue	0.5844	0.4448	1.4149	0.2346

Income reflects a binary coding where respondents are coded as “1” if reported income falls below the median income value.

Table 6:

F-test for Significance of Income (tertiary split) Interaction

Outcome	xti * Ii		Wti * Ii
	F	P-value	F	P-value
General Negative	0.4765	0.4902	0.0558	0.8133
Hostility	0.1849	0.6673	0.7407	0.3897
Guilt	0.2015	0.6536	0.3822	0.5366
Sadness	2.3467	0.1259	0.0274	0.8686
Fatigue	0.5756	0.4483	0.0556	0.8136

Income reflects a tertiary coding where respondents are coded as “−1” if reported income falls in the bottom third of the income distribution, “0” if reported income falls in the middle third of the income distribution, and as “1” if reported income falls in the top third of the income distribution.

When income is coded as a median split (Table 5), there are non-significant improvements in model fit for an interaction between educational debt stress and income, suggesting that income plays a negligible role in the relationship between educational debt stress and mental health fluctuations in a minority sample. Regarding “college stress”, we again find non-significant improvements in model fit for all mental health measurements. When income reflects a tertiary coding, or the income distribution is divided into thirds, (Table 6) there is once more no support for income moderation on educational debt stress or “college stress” across all five mood measurement scales. As “college stress” is about perception of the rigor of courses and anxiety regarding ones’ own future employment, we would not expect household income to have large effects on this measure. Taken in total, all the estimated models tell the same story: “college stress” is an important factor partially mediating the relationship between educational debt stress and mental health with little evidence of income moderation.

Finally, we replicated the income moderation analysis using a binary indicator for Latinx race/ethnicity, to assess if educational debt-stress and “college stress” operated differently for Latinx compared to African American students. Similar to income, race/ethnicity is a time-invariant variable with main effects already captured in the person fixed effects. As such, we added interaction effects between Latinx indicator and both educational debt stress and “college stress.” The results of these analyses, which are presented in Table 7 and 8, do not provide consistent evidence of differences in stress and “college stress” by race/ethnicity. Though our study is not well-suited (small N) to detect racial/ethnic differences in our key co-variates and outcomes, this analysis provides provisional evidence of similarities in reported educational debt stress and “college stress” for Latinx and Black students.

Table 7:

F-test for Significance of Latinx Interaction with Debt Stress

Outcome	xti * Ii
	bstress	bix_int	F	P-value
General Negative	0.282	0.038	8.977	0.003**
Hostility	0.196	0.114	0.408	0.372
Guilt	0.246	0.140	1.648	0.200
Sadness	0.204	0.192	0.020	0.886
Fatigue	0.220	0.095	1.934	0.165

Latinx reflects a binary coding where respondents are coded as “1” if reported race/ethnicity is “Latinx” and “0” if “Black” or “Other”. Table includes coefficients for non-Latinx student’s educational debt, b_stress, and Latinx student’s educational debt stress, b_{ix_int.} Latinx student’s educational debt stress expressed through an interaction term between educational debt stress and binary coding for reported Latinx race/ethnicity.

Table 8:

F-test for Significance of Latinx Interaction with “College Stress”

Outcome	wti * Li
	blwt	blx_int	F	P-value
General Negative	0.171	0.136	0.223	0.637
Hostility	0.099	0.146	0.329	0.566
Guilt	0.091	0.229	3.439	0.064
Sadness	0.108	0.027	0.999	0.318
Fatigue	0.256	0.217	0.236	0.627

Latinx reflects a binary coding where respondents are coded as “1” if reported race/ethnicity is “Latinx” and “0” if “Black” or “Other”. Table includes coefficients for non-Latinx student’s “College Stress” effect, b_lwt. and Latinx students “College stress” effect, b_{lx_int}. Latinx student’s “College Stress” expressed through an interaction term between “College Stress” and binary coding for reported Latinx race/ethnicity.

Robustness Checks

We also examined model assumptions and assessed the robustness of the presented findings. For example, we tested the validity of our “college stress” variable by removing the item associated with post-graduation job expectation, leaving our construct only with items related to courseload. When our base models were rerun, the findings were largely consistent. This signaled to us that beyond loading onto one factor, the three items do seem to cluster together to explain an underlying concept, what we deem “college stress”. Additionally, when BIC scores were generated to determine which model in our series provided the better parameterization across outcomes, “college stress” with all three items consistently provided the better fit.

Next, we sought to determine if psychological states were dependent over time. If so, our main models may overestimate the explanatory power of the designated predictor and the overall strength of the association. To assess if how students felt the previous day influenced reports of debt stress the following day, we estimated models with emotional scales as lagged predictors. We find that the inclusion of these lagged variables did not significantly change our results, but refrain from including them as our core results because lagged variables in a fixed effect context are vulnerable to nickell bias (Nickell 1981).

An additional concern was that repetitive questioning, a feature of daily diary designs, may have primed our respondents or increased their sensitivity to stress and debt-related questions. If this were true, reported levels of educational debt stress and components of “college stress” may be inflated as a consequence of study design. To test this theory, we plotted the average reported debt stress across days, finding that reported debt stress decreased over the study period. We also tested whether reported debt stress on any particular day was significantly different from reported debt stress on any other day. While reported mean debt stress on the first day of the study was statistically larger than reported debt stress on other days, when “Day 1” was removed from subsequent analyses results were not significantly different (as expected given the consistency of the lagged results).

A final concern was the issue of balanced data across our key covariates. Specifically, we suspected that the distribution of reported debt stress may be unbalanced such that most high-income students would not report high levels of debt stress. To test this hypothesis, we plotted debt stress as a function of income and demarcated the plot area by mean level of reported debt stress and income. Contrary to our concern, we find that students across the income distribution report both high and low levels of debt stress with a reasonably balanced joint distribution.

Discussion and Conclusion

This study employed a two-week daily diary design to assess the mental health consequences of educational debt-induced stress among a sample of predominately Black and Latinx students attending a large Midwestern research university. This research then considers facets of the college experience, “college stress” and time-use, as potential pathways by which educational debt stress may impact student mental health. We find that the effect of debt stress on mental health is influenced by “college stress” or the degree to which courseloads and post-graduation job expectation color the college experience while accruing debt. Together, the findings demonstrate that educational debt-induced stress is a robust predictor of general negative emotion, hostility, guilt, sadness, and fatigue at the daily level. Moreover, the relationship between educational debt stress and these five mood measurements was partially mediated by “college stress”, with time use and income offering minimal contributions. Below, we discuss study implications.

First, this study shows that stress incurred from thinking about educational debt is linked to negative mental health from day-to-day, thus supporting previous educational debt-mental health literature. While previous studies document that students with higher levels of financial concern tend to possess significantly worse mental health scores, this data often reflects cross-sectional analysis of measurements collected at one point in time (Walsemann, Gee, and Gentile 2015) or analysis of repeated measures collected on an annual or semesterly basis (Walsemann, Gee, and Gentile 2015; Cooke et al. 2004; Richardson et al. 2017). Leveraging dynamic daily diary data allowed this study to assess the day-to-day mental health burden of educational debt stress for currently enrolled students. The results suggest that the experience of educational debt is a strain associated with day-to-day fluctuations in students’ feelings of negative emotion, hostility, guilt, sadness, and fatigue.

Finding these results in a sample of predominately Black and Latinx students suggests that when students take on debt they may incur health costs that come due long before their monetary cost (Quadlin and Rudel 2015). Seamster and Charron-Chénier (2017) recently argued that while educational debt appears to provide marginalized individuals with opportunities for social and economic progress, in the long term it “reproduce[s] inequality and insecurity for some while allowing already-dominant social actors to derive significant profits” (Seamster and Charron-Chénier 2017: 200), a process they deem “predatory inclusion”. We build upon this work by suggesting that the consequences of “predatory inclusion” go beyond economic loss and can be identified in the short-term, or while students are enrolled in college. The persistence of negative feelings grounded in the day-to-day experience of educational debt may be a mechanism by which “predatory inclusion” shapes minority mental health.

Second, we find support for “college stress” as a contributor to the link between educational debt stress and mental health. Although the size of the effect of “college stress” on mental health may seem modest, the results remain after controls for observed and unobserved time-invariant characteristics are accounted for across all five affective mood outcome model series. Finding statistical significance for “college stress” has salient implications for the broader conversation on student loans, college experience, and degree completion. While scholars have found that debt can help and hinder student’s chances of graduation (Dwyer et al. 2012), its effects are not uniform within the population (Quadlin and Rudel 2016). For example, Black and low-income students are less likely to complete college if they accrue heavy debt in the first year (Kim 2007). Concerns and/or stress over affordability may directly influence student’s decision to complete college. Alternatively, debt accumulation may shape “the experiential core of college life” (Stevens et al. 2008, 131) in ways that make degree completion difficult, if not impossible, for some students. Future studies might consider if processes like “college stress” decrease the quality of the educational experience and contribute to drop-out rates. If so, students may be doubly disadvantaged in the long run by having accumulated debt without obtaining credentials.

As previously noted, the stress process model (SPM) emphasizes that while stress is universal, our respective locations within a racialized and classed social structure shape the intensity and range of stress outcomes that individuals manifest. Further, the stress process model underscores how different sources of stress may combine and interact with one another to influence individuals’ daily lives and health profiles. This study focuses on how educational debt stress and “college stress” work in tandem to shape the mental health of minority students. While all students are likely to experience some degree of stress from educational debt and/or college-related factors, we draw on literature supporting racialized differences in family socioeconomic status, reliance on student loan debt, and risk of loan default use (Jackson and Reynolds 2013; Ratcliffe and McKernan 2014; Cunningham and Santiago 2008), as well as racialized differences in acclimation to campus climate and institutional norms (Gross et al. 2009), to warrant our investigation of how Black and Latinx students uniquely experience educational debt and college stress.

While not investigated in this study, we acknowledge that predominately white college campuses are places where racial/ethnic minority students also encounter discrimination and vicarious racism (Cheadle et al. 2020; Jelsma et al. 2021; Jochman et al. 2019; Tynes et al. 2013; Solorzano et al. 2000). Future work might explore how students of color's more negative experiences in college based on their race/ethnicity interact with seemingly race-neutral stressors like educational debt and college stress to shape mental health.

Third, we find minimal support for an influence of time use on the relationship between educational debt stress and mental health. By this we mean that fluctuation in time spent in class, studying, working, or relaxing did not influence mental health among students in our models controlling for both person and time fixed effects. ICC (intraclass correlations) show considerable within-person variability across all 4 time use measures. At the same time, students report a relatively high average for time spent relaxing (10.89 hours). Taken together, time use variability may not be consistently associated with mental health fluctuations due to the considerable recovery time reported by students in our sample. In addition, although time use fluctuates, it is likely to be quite predictable from day to day. We therefore might expect associations to be magnified in populations where individuals have less recovery time and/or time use is less predictable.

Fourth, contrary to previous literature, this study finds no support for income moderation in the association between educational debt and student mental health. Specifically, the degree of association did not vary by level of estimated household income. The non-significant effect of income also holds true for “college stress” across all mental health dimensions. This null finding of income moderation emphasizes that “college stress” is a pathway from educational debt to mental health that is not primarily about affordability, but about the qualitative experience of college while accumulating debt.

Limitations

This study is not without limitations. StudentHD2 is a small convenience sample and replication is needed, as well as larger and more heterogeneous random samples to confidently identify effect sizes and increase generalizability. For instance, while we were able to provide provisional support for similarities in Black and Latinx students’ experience of debt stress and “college stress”, future studies with larger, random samples will be able to more confidently assess these relationships and explore heterogeneity within Black and Latinx populations. Further, because the daily diary study followed students for a maximum of two weeks, we could not examine the relationship between debt and mental health, as well as the mediating role of “college stress”, over longer time horizons to account for evolving identities and changing events over the college years. Previous work suggests that students become more debt tolerant as they progress through university (Lea et al. 2001). Given that our sample is comprised of majority 1^st and 2^nd year students (66%), we might expect an attenuated relationship between debt and mental health in an older sample of students. Moreover, if student’s attitude toward debt shifts as they come closer to graduation, it is highly likely that the impact of “college stress” on the relationship between student debt and mental health also shifts. Last, while there was little evidence of income moderation, conclusions must be drawn with care considering the difficulty children and young adults have in reporting total household income and wealth, which was not available in this study, but might be the more important contributor.

Despite these limitations, this study has a number of strengths. One of the key advantages of utilizing a daily diary design is that participants are able to report experiences shortly after they take place, tightly linking psychological states and social experiences in time. Thus, we are able to provide a detailed snapshot of the day-to-day college experience. This design also allows controlling for invariant factors, thereby adjusting for a large range of confounders. Furthermore, the study focuses on within-person variation among a predominately minority sample. In doing so, this study highlights the experiences of the individuals who are at a greater risk of both student borrowing and poor health. While sample size limited the analysis of ethnic variation, our inclusion of Latinx students provides a bases for future exploration on the relationship between debt and mental health in this understudied group. Additionally, we introduce “college stress” as a previously unconsidered pathway by which stress incurred from educational debt may impact mental health. This pathway may be particularly useful for understanding the quality of educational experiences and ultimately differential matriculation rates (Cook & Cordova, 2006; Kim 2007; Chen and Desjardins 2010). To this end, we hope that the presented findings not only inspire further empirical investigation but also prove beneficial to clinicians and policy makers in university settings.

Implications for Practice

Regarding recommendations for alleviating educational debt stress and college stress among students, particularly racial minorities, we might examine more closely and potentially model practices at Historically Black Colleges and Universities (HBCUs) and Hispanic-Serving Institutions (HSIs). For example, in 2014 and 2015 the Gallup-Purdue Index compared life outcomes for Black college graduates of HBCUs to outcomes for Black alumni of other colleges (Gallup 2015; Startz 2021). On average, HBCU grads had better outcomes on all assessed measures, with the advantage being largest for financial well-being, purpose well-being, and social well-being. When asked what happened in college that made such a difference, HBCU alumni reported more favorable experiences such as having professors that cared about them as people and feeling well integrated into academic and campus life.

Latinx students attending HSIs have also reported a number of experiential differences compared to their peers at Predominantly White Institutions (PWI). Nelson Laird and colleagues (2004) found Latinx students at HSIs reported a more supportive environment and greater levels of overall development compared to Latinx students at PWIs. Additionally, Latinx students at HSIs, compared to their peers at PWIs, identified more resources to increase their sense of belonging on campus and reported receiving additional peer support in academic areas (Musoba et al. 2013) and financial literacy (Smith 2019). Taken together, campus climates in which minority students feel included and supported, along with institutional policies to ensure regular communication regarding finances, may prove beneficial to student mental health and well-being.

Appendix

Table A-1:

Person and Time Fixed Effects Models for Educational Debt Stress and “College Stress” for Black and Latinx students only

	General Negative		Hostility		Guilt		Sadness		Fatigue
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
Educational Debt Stress	0.180*** (0.041)	0.118** (0.043)	0.162*** (0.046)	0.114* (0.049)	0.193*** (0.043)	0.136** (0.045)	0.216*** (0.044)	0.187*** (0.046)	0.164*** (0.045)	0.075 (0.046)
Labor Worth Tax		0.159*** (0.037)		0.119** (0.040)		0.144*** (0.038)		0.074 (0.038)		0.242*** (0.039)
Constant	0.073** (0.026)	0.068** (0.026)	0.031 (0.028)	0.026 (0.028)	0.078** (0.027)	0.073** (0.026)	0.029 (0.027)	0.026 (0.027)	0.003 (0.028)	−0.007 (0.027)
Fixed effects?	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N t	895	895	849	849	846	846	897	897	852	852
Adjusted R 2	0.520	0.530	0.478	0.483	0.544	0.552	0.481	0.483	0.472	0.496

^*p< .05.

^**p < .01.

^***p < .001.

Notes: Observations are drawn from a sample of 78 students. Black and Latinx students only; the 5 other students from different race/ethnic backgrounds are omitted.

Table A-2:

Income Distribution for Sample

	Freq.
Less than $10,000	4
$10,000 - $19,999	12
$20,000 - $29,999	10
$30,000 - $39,999	11
$40,000 - $49,999	7
$50,000 - $59,999	9
$60,000 - $69,999	5
$70,000 - $79,999	5
$80,000 - $89,999	6
$90,000 - $99,999	1
$100,000 - $149,999	7
More than $150,000	6