Unpacking the Antecedents and Consequences of Resilience to Socioeconomic Risks: Changes from Mid- to Late Adolescence

Abstract

Poverty-related risks exact detrimental effects on young people’s development, highlighting the need to understand the processes through which youth sustain resilience. This study examined the proximal contextual antecedents and developmental consequences of socioeconomic risk and resilience in middle (9^th grade) and late adolescence (12^th grade). Using survey data from a large longitudinal study (n = 738; 54% female; 42% White, 32% Latino/a/x, 12% biracial or multiracial, 8% Asian, and 6% Black), youth were classified as above or below the 75th percentile on socioeconomic risk and psychological resilience, yielding four risk and resilience groups: a) high-risk and high-resilience, b) low-risk and high-resilience, c) high-risk and low-resilience, and d) low-risk and low-resilience groups. In middle adolescence, the school and family contexts were particularly salient for distinguishing risk and resilience groups; however, by late adolescence, the family context remained the only consistent resource differentiating resilient youth. In middle adolescence, youth who were resilient in the face of economic stress had better mental and physical health than their less resilient peers. In contrast, socioeconomic risk regardless of resilience was problematic for academic achievement. In later adolescence, only the advantages in mental health persisted for youth resilient in the face of socioeconomic risk. Findings from the current study highlight potential targets in proximal contextual resources to improve supports for youth experiencing socioeconomic disadvantage.

In 2024, more than 10 million individuals under the age of 18 in the United States lived in poverty (Shrider & Bijou, 2025), and an additional 15 million children resided in low-income families (defined as 200% of the federal poverty line; Annie E. Casey Foundation, 2025). This represents approximately one-third of the child and adolescent population in the United States, and the socioeconomic disadvantage that these young people experience is a significant social stressor in their lives. Indeed, a host of scholarship has documented detriments tied to low socioeconomic status (SES) across all domains of adolescents’ development, from their physical and mental health to their educational success (Devenish et al., 2017; Jensen et al., 2017; Lawson et al., 2018; Letourneau et al., 2013; Sirin, 2005). These disadvantages persist and accumulate across time, ultimately heightening the risk for poorer life course outcomes, including lower occupational status and income, greater reliance on social safety nets, higher incarceration rates, and a greater morbidity and mortality in later life (Cohen et al., 2010; Link & Phelan, 1995; Ratcliffe, 2015).

Given the challenges of socioeconomic disadvantage throughout the life course, identifying young people who are resilient in the face of economic adversity is critical, as this can provide key insights for improving interventions and supports that can ensure all young people can thrive. Moreover, there is a well-established need for research that examines the proximal developmental contexts that promote thriving as well as how resilience experiences vary in relation to both developmental timing and domains of well-being (Chen et al., 2023; Masten et al., 2021; Ungar & Theron, 2020). The current study seeks to fill these gaps by investigating the contextual resources across families, schools, and peers that promote resilience as well as the impacts of resilience for youth’s mental health, physical health and health behaviors, and academic success. We do this with careful attention to developmental timing, focusing on potential variation from middle to late adolescence.

Resilience as a Developmental Construct

According to Masten and colleagues (2021), resilience is “the capacity of a dynamic system to adapt successfully through multisystem processes to challenges that threaten system function, survival, or development of the system” (p. 521). Conceptualizations of resilience draw from developmental models that suggest adaptation occurs both under the skin (via genetics and neurobiology) and in concert with, and in response to, the world around the individual, tapping into the micro- to macroecological contexts in which individuals live out their daily lives (Bronfenbrenner & Morris, 2006; Griffiths & Tabery, 2013). Theories operationalizing resilience highlight two components—a risk factor that places an individual at heightened probability for experiencing a negative developmental outcome and evidence of successful adaptation or recovery—that essentially capture thriving in the face of developmental threat (Chen et al., 2023; Masten et al., 2021; Mesman et al., 2021). Risk factors can be broad, spanning the structural (e.g., socioeconomic disadvantage, structural racism) to uncontrollable sociohistorical events (e.g., war, natural disasters) to more proximate (e.g., adverse childhood experiences, family conflict). There is recognized diversity in adaptation markers as well, with operationalization of adaptation encompassing general (e.g., self-control) to domain specific markers (e.g., social competence, educational success, lack of psychopathology); definitions of resilience generally allow for either successful adaption with no initial distress in the face of threat or successful recovery after initial distress tied to the adverse event (Anderson & Priebe, 2021).

There are, however, debates around some of the nuances of resilience as a developmental construct. For example, varying schools of thought position resilience as more trait-like, capturing an individual’s disposition that drives successful adaptation to threat and adversity, versus a characteristic that is more fluid, contextually dependent, and variable across developmental periods (Anderson & Priebe, 2021; Troy et al., 2023). Testing developmental models that explore resilience across time is key to unpacking whether resilience is indeed a more static trait or instead a skill that can be developed with experience and exposure to life course adversity. Additionally, many scholars suggest that resilience may be domain specific, such that thriving may not be consistent across developmental outcomes (Chen et al., 2023; Troy et al., 2023). Here specifically, scholars draw on the conceptualization of skin-deep resilience (Brody et al., 2013), wherein individuals may exhibit successful adaptation to threat and adversity in some domains (e.g., psychological well-being, academics), while showing signs of distress in others (e.g., physiological functioning; Chen et al., 2023; De France et al., 2022). The nascent scholarship on skin-deep resilience is limited but suggests that studies of resilience’s impacts for subsequent well-being should be broad reaching, considering individuals’ outcomes across developmental domains.

Resources that Promote Resilience in Adolescence

Given that a substantial proportion of youth will face threat in their young lives, understanding exactly how to promote thriving in the face of adversity is critical for equipping adolescents with the tools to foster well-being in the short and long term. Scholars (e.g., Hostinar & Miller, 2019; Mesman et al., 2021) have identified factors at both the intraindividual and interindividual levels that promote resilience, which Masten and colleagues (2021) have termed multisystem resilience factors. Internal factors include positive sense of self (e.g., self-efficacy, esteem), regulatory skills (e.g., self-control, self-regulation), and optimism and future orientation, while at the interpersonal level, promotive resilience factors include high-quality relationships with important others (e.g., family, friends, educators) and promotive proximal environments (e.g., neighborhood cohesion, positive school climates; Hostinar & Miller, 2019; Masten et al., 2021). In their theoretical model, Hostinar and Miller (2019) suggest that these factors can potentially inoculate the individual, essentially immunizing them from the stress response.

In the current study, we focus on the proximal contexts of adolescents’ lives, namely their families, peer groups, and schools, as the individuals in each represent important socializing and support agents during this developmental epoch. While extensive research has examined how proximal contexts influence self-reported resilience (e.g., bouncing back after difficulties, ease in making it through stressful life events), a much smaller research base has focused specifically on more objective resilience indicators that capture both stress exposure and adaptation. Although limited in scope, this literature has highlighted contextual factors that promote resilience that are in line with resilience theories (Hostinar & Miller, 2019; Masten et al., 2021). For example, family cohesion and high-quality parent-child relationships promoted resilience for youth exposed to maltreatment, high levels of violence, or war trauma (DiClemente et al., 2018; Eisman et al., 2015; Fayyad et al., 2017; Yoon et al., 2021), although there were some nuances related to whom these effects were more pronounced (e.g., boys vs girls). Likewise, school connectedness has been found to promote the resilience of youth exposed to economic adversity and intergenerational mental health problems (Mastromatteo et al., 2025; Oldfield et al., 2018). The scholarship on peer relations and resilience is more equivocal, with mixed-methods research highlighting the promotive effect of positive peer relationships for foster care-exposed youth (Strolin-Goltzman et al., 2016), but other research finding high levels of peer attachment were linked to less resilience (Oldfield et al., 2018). In the current study, we examine how aspects of the family, school, and peer ecological contexts can promote thriving in the face of economic adversity. Based on the stress-buffering hypothesis (Cohen & Wills, 1985), the current study also allows for examination of the benefit of social integration for all youth regardless of level of adversity.

Resilience to Socioeconomic Risk and Subsequent Well-being

Theoretical conceptualizations of resilience require both risk and adaptation, and a small body of research consistent with this operationalization have examined whether resilient youth exhibit consistent levels of positive well-being across developmental domains or if instead variation in outcomes is observed. These investigations are driven, in no small part, by theories around skin-deep resilience (Brody et al., 2013). Informed by the literature on high-effort coping symbolized by John Henryism (James, 1994), Brody and colleagues (2013) put forth the notion that resilience observed in youth experiencing adversity may be only skin level deep—that is, these thriving youth may display high levels of psychological well-being but, beneath the skin, their physiological functioning may be compromised. To test the potential tradeoffs of resilience, they identified resilience through considering socioeconomic adversity (risk factor) and self-control and scholastic and social competence (adaptation) in early adolescents, and they found that resilient youth (i.e., those who experiences high levels of socioeconomic adversity but displayed high levels of competence and control) had better mental health but poorer physical health in young adulthood (Brody et al., 2013).

These findings have been further explored in subsequent research, with general support for the skin-deep resilience theory observed. For example, Chen and colleagues (2023) examined youth with chronic illness ages 8 to 17 years of age, finding that resilient youth (i.e., those high in self-control/striving) displayed lower levels of both anxiety and depression but marginally worse health (i.e., higher proinflammatory phenotype). Likewise, other research, examining socioeconomic adversity (risk) and John Henryism high-effort coping (adaptation) have shown evidence for skin-deep resilience, such that low levels of SES were unrelated to youth’s mental health when John Henryism coping was high, but these youth had greater cardiometabolic risk (Ehrlich et al., 2024). The negative health impacts of resilience have also been replicated in research with White youth exposed to low income in late childhood in relation to poorer allostatic load in early adolescence and greater increases in allostatic load into young adulthood (De France et al., 2022). Likewise, Brody and colleagues (2016) observed that Black youth displaying resilience to socioeconomic disadvantage in late adolescence evinced fewer depressive symptoms in young adulthood but a greater likelihood of developing type 2 diabetes, but this pattern of skin-deep resilience was not observed with White youth. Taken together, this body of work generally supports the “cost of coping” hypothesis proposed by skin-deep resilience theory. While high-effort coping and self-control may allow economically disadvantaged youth to succeed academically and psychologically, the physiological toll of maintaining this effort under stress may compromise physical health. However, the majority of skin-deep resilience research has focused on static associations within a single developmental period rather than examining changes over time (with the exception of De France et al., 2022). This leaves a critical gap in understanding whether these trade-offs change during the transition from middle to late adolescence.

Current Study

The current study uses data from a longitudinal study with a community-based sample in the Southern United States to comprehensively examine the interpersonal antecedents and developmental consequences of resilience for youth during middle and late adolescence. Specifically, we sought to answer three research questions. First, we examined which interpersonal resources promoted resilience among adolescents exposed to socioeconomic risk. Here, we were specifically interested in the interpersonal assets that youth have in their proximal contexts, including parent trust, friend support and disclosure, and school climate. Based on prior research (DiClemente et al., 2018; Eisman et al., 2015; Fayyad et al., 2017; Yoon et al., 2021), we hypothesized that family and school resources would promote resilience in the face of SES-related adversity; given the equivocal nature of prior research on the impact of peer resources (Oldfield et al., 2018; Strolin-Goltzman et al., 2016), we offered no hypotheses as to the influence of friend support and disclosure on the likelihood of thriving in the face of socioeconomic disadvantage. Second, we investigated how socioeconomic risk and resilience came together to influence adolescents’ well-being across mental and physical health and academics. Consistent with theories of skin-deep resilience (Brody et al., 2013; Chen et al., 2023), we hypothesized that resilient youth would exhibit strong psychological well-being and academic success but poorer health than youth who experienced low socioeconomic disadvantage. Finally, we explored whether the pattern of relations across antecedents and consequences were similar during middle and late adolescence. Given the paucity of the extant research on risk and resilience that has captured both adversity and adaptation, we offered no specific hypotheses regarding potential developmental differences.

Using a longitudinal, community-based sample and harnessing multinomial logistic and multiple regression analyses across two developmental periods, the current study addresses several limitations of the existing research base on risk and resilience identified by scholars. This includes a comprehensive investigation of the multisystem antecedents of resilience (Chen et al., 2023; Ungar & Theron, 2020), spanning the proximal contexts and key socializing agents of adolescence, an examination of the consequences of resilience spanning developmental domains (i.e., educational, mental health, physical health; Hostinar & Miller, 2019), and an examination of developmental timing, focusing on these relations across adolescence (Ehrlich et al., 2024; Hostinar & Miller, 2019; Masten et al., 2021). As such, this work can provide key insights into how to best promote successful adaptation in the face of adversity and risk as well as identify the developmental supports necessary to foster the overall health and well-being of resilient youth.

Method

Participants and Procedures

Participants were drawn from an ongoing, longitudinal study, Project Pisces. The sample (N = 1,032) was recruited in eighth grade from middle schools in a large, metropolitan city in the Southwestern United States. Recruitment occurred in two cohorts during the 2016–2017 and 2017–2018 school years. Participants with parent consent and student assent for the longitudinal study component were invited to complete follow-up surveys each spring throughout high school. Of those eligible, 751 adolescents completed the Wave 2 assessment in ninth grade (87%). All study activities were approved by the University of Texas at Austin Institutional Review Board (# 2016-02-0149), and participants were compensated $25 for completing the survey.

The final analytic sample included only youth with valid risk (i.e., socioeconomic risk) and resilience data (i.e., psychological resilience) in middle and/or late adolescence (n = 738). For the analytic sample, 42% identified as White, 32% as Latino/a/x, 12% as biracial or multiracial, 8% as Asian, and 6% as Black. In total, 54% were female and 46% were male.

Measures

Socioeconomic Risk

Socioeconomic risk was assessed using six dichotomous indicators reflecting family structure, neighborhood context, and economic resources, consistent with prior work (Brody et al., 2013; Chen et al., 2023). All indicators were coded to reflect the presence of a given socioeconomic risk factor. Adolescents reported whether they lived with two biological parents (0 = resides in home with both biological parents, 1 = not residing in home with both biological parents). Two items assessed adolescents’ perceptions of neighborhood disadvantage: (a) the extent to which many families in their neighborhood were poor and (b) the extent to which many adults were unemployed (0 = strongly disagree to 3 = strongly agree). Responses were dichotomized such that agreement with each statement (i.e., agree or strongly agree; values ≥ 2) indicated neighborhood risk. Youth described their family’s social class by responding to the question, “How would YOU describe the wealth of your family?” (0 = very poor to 5 = upper-class/rich). Responses of 0 (very poor) or 1 (poor) were coded as 1, indicating economic hardship. Parent education was coded as a dichotomous variable (0 = any education beyond high school; 1 = no high school diploma or GED). Family economic disadvantage was captured through public school records based on eligibility for the federal Free or Reduced-price Lunch (FRPL) Program (0 = nonpoor, 1 = poor).

The six indicators were summed to create a composite count of socioeconomic risk (range: 0–6) separately for Wave 2 (middle adolescence) and Wave 6 (late adolescence). The composite variable was then dichotomized using the 75th percentile as a cutoff (0 = fewer than two risk factors, 1 = two or more risk factors).

Psychological Resilience

Psychological resilience was used to tap into youth adaptation and was measured using a mean composite of four items assessing how good youth felt they were at skills such as “Figuring out problems and planning how to solve them” and “Bouncing back from bad experiences.” Ratings for each item ranged from 0 (very bad) to 4 (excellent). The measure exhibited strong reliability (Cronbach’s α = .79 at both waves). A dichotomous indicator was created from the mean composite using the 75th percentile as a cutoff. Separate dichotomous indicators of adaptation were created for Wave 2 and Wave 6.

Interpersonal Resources

Adolescents’ interpersonal resources were measured at Waves 1 and 5.

School Belonging.

To measure school belonging (Johnson et al., 2001), adolescents indicated the extent to which they agreed with five statements (e.g., “I feel like I am a part of this school”) using a scale ranging from 0 (no way) to 4 (for sure yes). A composite mean score was created, with higher scores denoting greater school belonging (α_W1 = .78; α_W5 = .82).

School Peer Climate.

Adolescents’ perceptions of school peer climate were assessed using a modified version of the Inventory of School Climate – Student (ISC-S; Brand et al., 2003). Adolescents rated 10 items (e.g., “Students at this school are very interested in getting to know other”) on a scale ranging from 0 (no way) to 4 (for sure yes). A mean composite score was created, with higher scores representing a more positive school peer climate (α_W1 = .80; α_W5 = .80).

Parent Trust.

Parent trust was assessed using six items from Armsden and Greenberg’s Inventory of Parent and Peer Attachment (1987; e.g., “My parents respect my feelings”). The scale ranged from 0 (never) to 4 (always). A mean composite was created, with higher scores denoting greater parent trust (α_W1 = .92; α_W5 = .93).

Friend Support.

Friend support was assessed using five items from the friend trust subscale of the Inventory of Parent and Peer Attachment (Armsden & Greenberg, 1987; e.g., “My friends respect my feelings”). The rating scale ranged from 0 (never) to 4 (always). A mean composite was created, with higher scores indicating greater friend support (α_W1 = .91; α_W5 = .93).

Disclosure to Friends.

Friend disclosure was assessed using five items from the Network of Relationships Questionnaire (Furman & Buhrmester, 2010; e.g., “Share secrets and private feelings”). Adolescents provided ratings on a scale from 0 (never) to 4 (always). A mean composite was created, with higher scores reflecting greater disclosure to friends (α_W1 = .88; α_W5 = .88).

Adolescents’ Well-being

Adolescents’ well-being indicators were measured at Waves 2 and 6.

Depressive Symptoms.

Adolescents completed the abbreviated 10-item Center for Epidemiologic Studies Depression Scale (CES-D; Andresen et al., 1994). Adolescents reported how often they had experienced various depressive symptoms in the past week, such as feeling lonely, fearful, and depressed. Responses varied from 0 (rarely or none of the time (Less than 1 day)) to 3 (Most or all of the time (5–7 days)). A mean composite was generated, where higher scores denoted greater depressive symptoms (α_W2 = .80; α_W6 = .83).

Self-Rated Health.

Self-rated health was evaluated using a single item asking, “In general how would you rate your health?” (Harris, 2018). Ratings ranged from 0 (poor) to 4 (excellent).

Sleep.

Sleep disturbances were assessed using two items (Harris, 2018): how often adolescents had trouble falling and staying asleep over the past four weeks. The rating scale ranged from 0 (never) to 4 (five or more times per week). A mean composite was created, with higher scores reflecting more frequent sleep disturbances (r_W2 = .61; r_W6 = .61).

GPA.

Grades in core content and elective courses were drawn from students’ school records. For each course, grades ranged from 0 to 100, and a mean score captured average grades in all courses across the fall and spring semesters of 9^th grade and 12^th grade.

Covariates

Student-level covariates included students’ gender (1 = female, 0 = male) and race/ethnicity (dummy codes for Latino/a/x and other race [African American/Black, Asian American, and biracial/other] with White as the omitted racial/ethnic group).

Analysis Plan

To examine the first research question (interpersonal resources promoting resilience in the face of socioeconomic risk), we first characterized joint patterns of socioeconomic risk and psychological resilience using cross-tabulations. This approach yielded a four-category typology, constructed separately for middle and late adolescence: (a) high-risk and high-resilience, (2) low-risk and high-resilience, (3) high-risk and low-resilience, and (4) low-risk and low-resilience groups. We then estimated a series of multinomial logistic regressions predicting adolescents’ membership in these risk/resilience groups from interpersonal resources. Models were estimated separately for middle and late adolescence and controlled for gender and race/ethnicity. The low-risk, high-resilience group was retained as a normative reference group, providing a meaningful comparison point against which the adaptation of higher-risk adolescents could be interpreted. To examine all pairwise comparisons, we rotated the reference group across models. To test our second research question (risk and resilience groups and adolescents’ subsequent well-being), we ran a series of multiple regression models, regressing the well-being outcomes on the risk/resilience groups and covariates, rotating the reference group to obtain all pairwise comparisons.

To capture potential variation in the pattern of relations in middle versus late adolescence, we compared nested models. In the first set of models, we included the relations between interpersonal resources and risk/resilience groups for both middle and late adolescence simultaneously. In the second set, the relations between risk/resilience groups and well-being in middle and late adolescence were modeled simultaneously. For each set of models, we first estimated unconstrained models where all relations were free to vary, followed by a second set of models that constrained central model pathways to be equal in middle and late adolescence. A scaled log-likelihood difference test determined whether the constraints led to a significant decline in model fit (Satorra & Bentler, 2010).

Analyses were conducted using Mplus Version 8.7 (Muthén & Muthén, 1998–2017). Missing data were handled using full-information maximum likelihood (FIML), enabling us to include all available data in the analyses. In examining missing data patterns, in middle adolescence, 570 participants (77.6%) had complete data and 165 (22.4%) were missing at least one variable. Relative to complete cases, those with missing data were more likely to be Latino/a/x (46.7% vs. 28.1%; χ² (1) = 19.41, p < .001) and less often from another racial group (Asian, Black, or biracial; 17.6% vs. 28.8%; χ² (1, N = 735) = 7.72, p = .005). They also had higher socioeconomic risk (t (733) = 3.32, p = .001), lower resilience (t (733) = 2.47, p = .014, poorer physical health (t (730) = 2.61, p = .009), and lower friend disclosure (t (699) = 2.69, p = .007). In late adolescence, 201 participants (65.5%) had complete data and 106 (34.5%) were missing at least one variable. Those with missing data were again more likely to be Latino/a/x (34.9% vs. 23.4%; χ² (1, N = 307) = 4.07, p = .044) and reported higher socioeconomic risk (t (305) = 2.17, p = .031). While FIML assumes data are missing at random (MAR, Enders, 2022), FIML is one of the most effective approaches for handling missing data even under some MNAR conditions (Tang & Tong, 2025). We utilized maximum likelihood with robust standard errors as an estimator to all models due to data non-normality.

Results

We began by examining the distribution of adolescents across the four risk and resilience groups in both middle and late adolescence. During middle adolescence, the largest proportion of youth fell into the low-risk, low-resilience group (41.2%), followed by the low-risk, high-resilience group (28.4%). The remaining adolescents were classified as high-risk, low-resilience (20.7%) and high-risk, high-resilience (9.5%). A comparable distribution was observed in late adolescence: 43.6% low-risk, low-resilience, 36.5% low-risk, high-resilience, 13.0% high-risk, low-resilience, and 6.8% high-risk, high-resilience. Among youth with data at both waves, 61.2% remained in the same risk–resilience group from middle to late adolescence, with the remainder moving groups across these developmental periods (detailed classification presented in Table S1). Descriptive statistics and bivariate correlations among the primary constructs of interest are presented in Table S2. Table S3 presents comparisons of central study constructs across the four risk-resilience groups.

Interpersonal Resources Associated with Risk and Resilience

Middle Adolescence

In middle adolescence, we found that interpersonal resources in the school and family contexts were particularly impactful (see odds ratios in Table 1; regression coefficients in Table S4). In the school context, higher school belonging decreased the odds of being in the low-risk and low-resilience group compared to both the high-risk and high-resilience group (OR = 0.59, [CI: 0.37, 0.95], p = .031) and the low-risk and high-resilience group (OR = 0.61, [CI: 0.45, 0.83], p = .002). School belonging, however, did not differentiate between any other risk and resilience groups. School peer climate distinguished high-risk youth from their low-risk peers. More specifically, more positive peer climate was associated with greater odds of membership in the low-risk and high-resilience (OR = 2.11, [CI: 1.16, 3.83], p = .014) and low-risk and low-resilience groups (OR = 2.14, [CI: 1.18, 3.87], p = .012) compared to the high-risk and high-resilience group. In addition, more positive peer school climate was associated with a lower likelihood of being in the high-risk and low-resilience group compared to the low-risk and high-resilience group (OR = 0.47, [CI: 0.28, 0.79], p = .005) and greater likelihood of being in the low-risk and low-resilience groups (OR = 2.16, [CI: 1.31, 3.54], p = .002) compared to the high-risk and low-resilience group. The two low-risk groups did not differ from one another in terms of peer school climate.

Table 1.

Likelihood of Membership in Risk-and-Resilience Groups by Predictors.

	LL vs. HH	LH vs. HH	HL vs. HH	LL vs. HL	LL vs. LH	HL vs. LH
	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	Group Comparison
Middle Adolescence (n = 735)
School Belonging	0.59* [0.37, 0.95]	0.97 [0.58, 1.60]	0.70 [0.43, 1.16]	0.84 [0.59, 1.20]	0.61** [0.45, 0.83]	0.73 [0.49, 1.08]	HH>LL; HH=LH; HH=HL; LH=HL; HL=LL; LH>LL
School Peer Climate	2.14* [1.18, 3.87]	2.11* [1.16, 3.83]	0.99 [0.54, 1.83]	2.16** [1.31, 3.54]	1.02 [0.68, 1.51]	0.47** [0.28, 0.79]	HH<LL; HH<LH; HH=HL; LH>HL; HL<LL; LH=LL
Parent Trust	0.76 [0.52, 1.10]	1.15 [0.76, 1.72]	0.71 [0.48, 1.05]	1.07 [0.82, 1.39]	0.66** [0.51, 0.85]	0.62** [0.45, 0.85]	HH=LL; HH=LH; HH=HL; LH>HL; HL=LL; LH>LL
Friend Support	1.53 [0.94, 2.47]	1.81* [1.05, 3.14]	1.51 [0.93, 2.44]	1.01 [0.72, 1.41]	0.84 [0.58, 1.22]	0.83 [0.54, 1.29]	HH=LL; HH<LH; HH=HL; LH=HL; HL=LL; LH=LL
Disclosure to Friends	1.02 [0.65, 1.61]	0.96 [0.60, 1.53]	0.92 [0.58, 1.47]	1.11 [0.82, 1.51]	1.07 [0.81, 1.41]	0.96 [0.68, 1.37]	HH=LL; HH=LH; HH=HL; LH=HL; HL=LL; LH=LL
Late Adolescence (n = 307)
School Belonging	1.11 [0.54, 2.29]	1.40 [0.68, 2.89]	1.05 [0.46, 2.40]	1.06 [0.62, 1.81]	0.80 [0.55, 1.15]	0.75 [0.43, 1.32]	HH=LL; HH=LH; HH=HL; LH=HL; HL=LL; LH=LL
School Peer Climate	2.42 [0.78, 7.57]	2.55 [0.79, 8.15]	2.18 [0.64, 7.46]	1.11 [0.51, 2.43]	0.95 [0.54, 1.70]	0.86 [0.37, 1.97]	HH=LL; HH=LH; HH=HL; LH=HL; HL=LL; LH=LL
Parent Trust	0.57 [0.26, 1.26]	0.71 [0.32, 1.56]	0.35* [0.14, 0.91]	1.61 [0.90, 2.87]	0.81 [0.58, 1.11]	0.50* [0.28, 0.90]	HH=LL; HH=LH; HH>HL; LH>HL; HL=LL; LH=LL
Friend Support	0.85 [0.36, 2.02]	0.89 [0.36, 2.20]	0.66 [0.25, 1.70]	1.30 [0.73, 2.31]	0.95 [0.60, 1.51]	0.73 [0.36, 1.49]	HH=LL; HH=LH; HH=HL; LH=HL; HL=LL; LH=LL
Disclosure to Friends	1.06 [0.59, 1.91]	1.00 [0.53, 1.87]	1.45 [0.69, 3.01]	0.73 [0.44, 1.24]	1.06 [0.70, 1.61]	1.45 [0.77, 2.74]	HH=LL; HH=LH; HH=HL; LH=HL; HL=LL; LH=LL

Note. Values are odds ratios (OR) with 95% confidence intervals (CI) in brackets. Middle adolescence: n = 735 (HH n = 70; LH n = 209; HL n = 153; LL n = 303). Late adolescence: n = 307 (HH n = 21; LH n = 112; HL n = 40; LL n = 134). The two developmental periods were modeled separately; 304 adolescents contributed to both, with a combined sample of n = 738. Within each period, models were estimated with FIML, so all cases contributed to estimation. HH = High-Risk & High-Resilience; LH = Low-Risk & High-Resilience; HL = High-Risk & Low-Resilience; LL = Low-Risk & Low-Resilience. The Group Comparison column summarizes the significant differences, where > and < identify the group reporting the higher level of the resources and = denotes no significant difference. Covariate effects are presented in Table S4.

^*p < .05.

^**p < .01.

^***p < .001.

When examining the family context, parent trust differentiated some of the risk and resilience groups. Higher parent trust decreased the odds of being in either the high-risk and low-resilience groups (OR = 0.62, [CI: 0.45, 0.85], p = .003) or the low-risk and low-resilience group (OR = 0.66, [CI: 0.51, 0.85], p = .002) compared to the low-risk and high-resilience group. Parent trust did not distinguish adolescents in the high-risk and high-resilience group from any of the other groups.

In the peer context, we observed minimal associations with peer support and no associations with peer disclosure. Notably, adolescents with more supportive friendships were more likely to be in the low-risk and high-resilience group than those in the high-risk and high-resilience group (OR = 1.81, [CI: 1.05, 3.14], p = .034).

Late Adolescence

In late adolescence, neither the school (i.e., school belonging, school peer climate) nor peer context (i.e., peer support, peer disclosure) distinguished among the risk and resilience groups (see Table 1 and S4). Parent trust, however, remained a key differentiator. Adolescents who reported significantly higher trust had decreased odds of being in the high-risk and low-resilience group than the high-risk and high-resilience (OR = 0.35, [CI: 0.14, 0.91], p = .030) and low-risk and high-resilience groups (OR = 0.50, [CI: 0.28, 0.90], p = .020). No additional group differences related to parent trust emerged.

Taken together, these findings indicate that school and family were the primary interpersonal processes distinguishing socioeconomic risk and resilience groups during middle adolescence, whereas by late adolescence, parent trust was the only consistent interpersonal resource differentiating resilient from non-resilient youth under higher socioeconomic risk.

Risk and Resilience and Adolescent Well-being

Middle Adolescence

We next examined the extent to which adolescents’ socioeconomic risk and resilience were associated with well-being in middle and late adolescence (see Table 2 and Table S5). When considering adolescents’ depressive symptoms, those in the high-risk and low-resilience (β = 0.24, p < .001) and low-risk and low-resilience groups (β = 0.25, p < .001) reported more depressive symptoms than those in the high-risk and high-resilience group. In addition, both high-risk and low-resilience adolescents (β = 0.28, p < .001) and low-risk and low-resilience adolescents (β = 0.30, p < .001) reported greater depressive symptoms than those in the low-risk and high-resilience group.

Table 2.

The Associations Between Risk/Resilience Groups and Outcomes.

	Middle Adolescence (n = 735)				Late Adolescence (n = 307)
	Depressive Symptoms	Health	Sleep	GPA	Depressive Symptoms	Health	Sleep	GPA
	β (SE) [95% CI]	β (SE) [95% CI]	β (SE) [95% CI]	β (SE) [95% CI]	β (SE) [95% CI]	β (SE) [95% CI]	β (SE) [95% CI]	β (SE) [95% CI]
Model (vs. High-Risk & High-Resilience)
High-Risk & Low-Resilience	0.24*** (0.05) [0.14, 0.33]	−0.20*** (0.05) [−0.30, −0.10]	0.09 (0.06) [−0.02, 0.20]	−0.01 (0.07) [−0.15, 0.13]	0.21* (0.09) [0.04, 0.38]	−0.17 (0.10) [−0.36, 0.02]	0.15 (0.09) [−0.03, 0.32]	0.07 (0.16) [−0.25, 0.39]
Low-Risk & High-Resilience	−0.05 (0.05) [−0.14, 0.05]	0.11 (0.06) [−0.00, 0.21]	−0.05 (0.06) [−0.16, 0.07]	0.43*** (0.07) [0.30, 0.56]	−0.03 (0.11) [−0.25, 0.19]	0.13 (0.14) [−0.13, 0.40]	−0.08 (0.12) [−0.32, 0.16]	0.68*** (0.16) [0.38, 0.98]
Low-Risk & Low-Resilience	0.25*** (0.05) [0.15, 0.36]	−0.17** (0.06) [−0.28, −0.06]	0.12 (0.06) [−0.01, 0.24]	0.37*** (0.07) [0.23, 0.51]	0.28* (0.12) [0.05, 0.51]	−0.11 (0.14) [−0.38, 0.16]	0.01 (0.12) [−0.24, 0.25]	0.65*** (0.16) [0.34, 0.96]
Model (vs. High-Risk & Low-Resilience)
High-Risk & High-Resilience	−0.17*** (0.04) [−0.24, −0.10]	0.15*** (0.04) [0.07, 0.22]	−0.07 (0.04) [−0.15, 0.02]	0.01 (0.05) [−0.10, 0.11]	−0.16* (0.07) [−0.29, −0.03]	0.13 (0.07) [−0.02, 0.27]	−0.11 (0.07) [−0.24, 0.02]	−0.06 (0.12) [−0.30, 0.18]
Low-Risk & High-Resilience	−0.31*** (0.05) [−0.40, −0.22]	0.33*** (0.05) [0.23, 0.43]	−0.15** (0.05) [−0.24, −0.05]	0.44*** (0.05) [0.33, 0.54]	−0.33*** (0.09) [−0.51, −0.16]	0.38*** (0.08) [0.22, 0.53]	−0.29** (0.09) [−0.46, −0.12]	0.57*** (0.13) [0.31, 0.84]
Low-Risk & Low-Resilience	−0.03 (0.05) [−0.14, 0.07]	0.08 (0.05) [−0.02, 0.18]	0.01 (0.05) [−0.09, 0.11]	0.38*** (0.06) [0.28, 0.49]	−0.03 (0.09) [−0.21, 0.15]	0.14 (0.08) [−0.02, 0.30]	−0.21* (0.09) [−0.38, −0.04]	0.54*** (0.14) [0.28, 0.81]
Model (vs. Low-Risk & High-Resilience)
High-Risk & High-Resilience	0.03 (0.03) [−0.03, 0.09]	−0.07 (0.04) [−0.14, 0.00]	0.03 (0.04) [−0.05, 0.11]	−0.28*** (0.05) [−0.37, −0.19]	0.02 (0.06) [−0.10, 0.13]	−0.07 (0.07) [−0.21, 0.07]	0.04 (0.06) [−0.09, 0.17]	−0.36*** (0.09) [−0.52, −0.19]
High-Risk & Low-Resilience	0.28*** (0.04) [0.19, 0.36]	−0.30*** (0.05) [−0.39, −0.21]	0.13** (0.05) [0.04, 0.22]	−0.39*** (0.05) [−0.49, −0.30]	0.23*** (0.06) [0.11, 0.36]	−0.26*** (0.06) [−0.38, −0.15]	0.20** (0.06) [0.08, 0.33]	−0.40*** (0.09) [−0.59, −0.22]
Low-Risk & Low-Resilience	0.30*** (0.04) [0.23, 0.38]	−0.28*** (0.04) [−0.36, −0.20]	0.17*** (0.04) [0.08, 0.25]	−0.10* (0.04) [−0.17, −0.02]	0.31*** (0.06) [0.20, 0.42]	−0.25*** (0.06) [−0.37, −0.13]	0.09 (0.06) [−0.04, 0.21]	−0.05 (0.05) [−0.14, 0.05]
Model (vs. Low-Risk & Low-Resilience)
High-Risk & High-Resilience	−0.15*** (0.03) [−0.21, −0.09]	0.10** (0.03) [0.03, 0.17]	−0.07 (0.04) [−0.14, 0.01]	−0.22*** (0.04) [−0.31, −0.14]	−0.14* (0.06) [−0.26, −0.02]	0.06 (0.07) [−0.08, 0.19]	−0.00 (0.06) [−0.13, 0.12]	−0.33*** (0.08) [−0.50, −0.17]
High-Risk & Low-Resilience	0.03 (0.04) [−0.06, 0.11]	−0.06 (0.04) [−0.15, 0.02]	−0.01 (0.04) [−0.09, 0.08]	−0.32*** (0.05) [−0.41, −0.23]	0.02 (0.06) [−0.10, 0.15]	−0.10 (0.06) [−0.20, 0.01]	0.14* (0.06) [0.03, 0.26]	−0.37*** (0.10) [−0.56, −0.18]
Low-Risk & High-Resilience	−0.28*** (0.04) [−0.35, −0.21]	0.26*** (0.04) [0.19, 0.33]	−0.15*** (0.04) [−0.23, −0.08]	0.09* (0.03) [0.02, 0.15]	−0.30*** (0.06) [−0.41, −0.20]	0.24*** (0.06) [0.13, 0.36]	−0.09 (0.06) [−0.21, 0.04]	0.05 (0.05) [−0.04, 0.14]

Note. Values are standardized regression coefficients (β) with standard errors in parentheses and 95% confidence intervals in brackets. Middle adolescence: n = 735 (high-risk & high-resilience n = 70; low-risk & high-resilience n = 209; high-risk & low-resilience n = 153; low-risk & low-resilience n = 303). Late adolescence: n = 307 (high-risk & high-resilience n = 21; low-risk & high-resilience n = 112; high-risk & low-resilience n = 40; low-risk & low-resilience n = 134). The two developmental periods were modeled separately; 304 adolescents contributed to both, with a combined sample of n = 738. Within each period, models were estimated with FIML, so all cases contributed to estimation. Covariate effects are presented in Table S5.

^*p < .05.

^**p < .01.

^***p < .001.

For health and health behaviors, adolescents in the high-risk and low-resilience group reported poorer self-rated health than those in both the high-risk and high-resilience (β = −0.20, p < .001) and low-risk and high-resilience groups (β = −0.30, p < .001), and adolescents in the low-risk and low-resilience group also reported poorer health than high-risk and high-resilience youth (β = −0.17, p < .01). Adolescents in the high-risk and low-resilience group (β = 0.13, p < .01) and the low-risk and low-resilience group (β = 0.17, p < .001) reported more sleep disturbances than those in the low-risk and high-resilience group.

Finally, when examining academic performance, adolescents in the low-risk and high-resilience (β = 0.43, p < .001) and low-risk and low-resilience groups (β = 0.37, p < .001) reported higher GPAs than those in the high-risk and high-resilience group. In addition, both high risk–low resilience (β = −0.39, p < .001) and low risk–low resilience adolescents (β = −0.10, p = .02) earned lower grades than their low risk–high resilience peers. Overall, the findings suggest that in middle adolescence, resilience seems to confer more advantages in relation to both mental and physical health, whereas socioeconomic advantage plays a central role in adolescents’ academic success.

Late Adolescence

In late adolescence, the results revealed meaningful distinctions across the risk and resilience groups. Consistent with patterns observed in middle adolescence, adolescents in the high-risk and low-resilience (β = 0.21, p = .017) and in the low-risk and low-resilience groups (β = 0.28, p = .016) reported more depressive symptoms than those in the high-risk and high-resilience group. In addition, adolescents in the high-risk and low-resilience group reported greater depressive symptoms than those in the low-risk and high-resilience group (β = 0.23, p < .001).

For self-rated health, youth with lower resilience reported poorer perceived health than their peers with higher resilience, consistent with observed patterns in middle adolescence. Specifically, compared to the low-risk and high-resilience group, adolescents in the high-risk and low-resilience group (β = −0.26, p < .001) and low-risk and low-resilience group (β = −0.25, p < .001) reported poorer self-rated health. No other differences emerged. For sleep, adolescents in the high-risk and low-resilience group reported more sleep disturbances than both the low-risk and high-resilience group (β = 0.20, p < .01) and the low-risk and low-resilience group (β = 0.14, p = .02), whereas sleep did not differ between the two low-risk groups, a pattern not observed earlier in adolescence.

Finally, youth who experienced higher socioeconomic risk reported worse academic outcomes than their lower risk peers, consistent with findings in middle adolescence. For academic performance, adolescents in the low-risk and high-resilience group (β = 0.68, p < .001) and those in the low-risk and low-resilience group (β = 0.65, p < .001) earned higher grades in school than adolescents in the high-risk and high-resilience group. Additionally, adolescents in the high-risk and low-resilience group reported lower grades than both the low-risk, high-resilience group (β = −0.40, p < .001) and the low-risk, low-resilience group (β = −0.37, p < .001). Overall, the pattern of effects observed in late adolescence suggest that socioeconomic advantage is strongly linked to academic performance, whereas resilience plays a greater role for mental and physical health.

Developmental Differences in Middle versus Late Adolescence

Scaled log-likelihood difference tests assessed whether the antecedents and consequences of risk–resilience group membership varied across middle and late adolescence. In investigating antecedent interpersonal resources, no significant differences emerged across developmental periods (TRd (15) = 13.83, p = .54). Likewise, we observed no developmental differences from middle to late adolescence for the models integrating the mental health, physical health and health behaviors, and academic outcomes (TRd (12) = 9.77–18.83, p = .09–.64).

Supplemental Analyses

Prior studies examining the well-being of resilient youth often utilize a different analytic strategy to capture resilience than that used in the current study (e.g., Brody et al., 2013). While our strategy was implemented to enable exploration of the interpersonal assets that promote resilience in the face of adversity, we conducted supplemental moderation analyses to examine whether psychological resilience moderated associations between SES risk and adolescents’ well-being, consistent with prior risk and resilience research. For each developmental period, Model 1 regressed the well-being outcomes on the continuous SES risk measure, Model 2 added the continuous psychological resilience measure, and Model 3 integrated the SES risk × psychological resilience interaction term. The middle and late adolescence models were fully saturated, and as such, fit statistics were not available.

Middle Adolescence

We first investigated how continuous SES risk was associated with depressive symptoms, self-rated health, sleep disturbances, and grades. As shown in Table S6 (Model 1), in middle adolescence, greater SES risk was significantly associated with higher depressive symptoms (b = 0.06, p = .002), poorer self-rated health (b = −0.16, p < .001), and lower grades in school (b = −2.72, p < .001). Model 2 indicated that higher psychological resilience was significantly associated with lower depressive symptoms (b = −0.30, p < .001), better self-rated health (b = 0.52, p < .001), fewer sleep disturbances (b = −0.35, p < .001), and higher grades (b = 1.14, p = .005; see Table S7). Model 3 integrated interaction effects between SES risk and psychological resilience; however, no significant interactions were observed (See Table S8).

Late Adolescence

As shown in Table S6 (Model 1), in late adolescence, greater SES risk was significantly associated with poorer self-rated health (b = −0.23, p < .001), more sleep problems (b = 0.19, p = .008), and lower grades (b = −3.07, p < .001). Model 2 indicated that higher psychological resilience was significantly associated with lower depressive symptoms (b = −0.33, p < .001) and better self-rated health (b = 0.53, p < .001; (See Table S7). Model 3 showed no significant interactions (See Table S8). Thus, these exploratory models did not provide evidence that psychological resilience buffered associations between SES risk and well-being when SES risk and psychological resilience are modeled as continuous variables, suggesting that these associations might be nonlinear and/or that the effects observed in our primary analyses were driven by consideration of individuals at the extremes (e.g., 75^th percentile).

Discussion

Informed by theories of risk and resilience (Masten et al., 2021), the current study sought to expand on the extant literature by examining both the antecedents and consequences of resilience in the face of socioeconomic disadvantage in middle and late adolescence. Aligning closely with operationalizations of resilience (e.g., Brody et al., 2013)—namely in integrating both adversity and adaptation factors—we examined how youth fell into risk and resilience patterns based on their socioeconomic risk and their psychological resilience. For the current sample, a substantial majority of youth were in groups characterized by low economic risk (69% in middle adolescence and 80% in late adolescence), with more youth with low economic risk reporting low levels of psychological resilience. In contrast, much fewer youth faced economic adversity, and of these, more also displayed low levels of psychological resilience, placing them at a double disadvantage. Resilient youth—that is those who experienced economic adversity but reported high levels of psychological resilience—were most rare.

Proximal developmental contexts differentiated across the risk and resilience groups. In middle adolescence, school belonging promoted resilience for both those facing economic adversity and those in more advantaged families. School belonging has been found to promote well-being across developmental domains (Allen et al., 2022a), suggesting it could be a malleable target for intervention. While meta-analytic research has highlighted the role of teacher support and youth’s internal assets in fostering greater school belonging (Allen et al., 2018), interventions designed to promote school belonging are rare, and their impacts are variable (Allen et al., 2022b). The findings reported here provide further evidence of the need for such interventions, particularly in middle adolescence.

Positive relationships with parents were also promotive of resilience. In middle adolescence, parent trust was particularly potent for those not facing economic adversity, whereas by late adolescence, parent trust also emerged as promotive of resilience for those facing economic adversity. Moreover, the positive effects of families were the only salient predictors of risk and resilience groups in both middle and late adolescence. Families are important socializing agents across the early life course, and the findings observed in the current study are consistent with research on other aspects of parent-child relationships—namely family cohesion and high-quality parent-child relationships—which have been found to promote resilience among youth facing adversity (DiClemente et al., 2018; Eisman et al., 2015). Unexpectedly, poorer experiences in the peer group, whether operationalized as overall peer school climate or friend support, were linked to a greater likelihood of being in the high risk and high resilience group. It may be that youth exposed to greater adversity are more likely to encounter risky peers or have interactions with peers that are more problematic for building resilience, such as greater co-rumination (Dishion & Tipsord, 2011). More research is needed to unpack the impact of peers, and qualitative inquiry may be particularly useful in this regard.

Moving to the consequences of risk and resilience for well-being, we found that youth who were resilient in the face of economic adversity displayed fewer depressive symptoms and better self-rated health in middle adolescence than those reporting low resilience (with and without economic adversity), and the differences for depressive symptoms but not health persisted into late adolescence. Theories of skin-deep resilience suggest that resilience can take a toll on youth facing substantial adversity (Brody et al., 2013; Chen et al., 2023). Our findings, however, offer limited support for this view, as resilient youth did not report poorer physical health or sleep than their resilient but low risk peers at either developmental period. Although their health advantage attenuated from middle to late adolescence, resilience was not accompanied by the physiological cost that skin-deep resilience would predict. Whether such a cost emerges later in development remains unclear, and thus longitudinal studies spanning adolescence into young adulthood may be particularly useful in tracking the accumulating disadvantages for health that resilience in the face of adversity may exact. In addition, skin-deep resilience models also focus on biological markers (e.g., C-reactive protein, allostatic load; Brody et al., 2013; Chen et al., 2020), and it is possible that these markers might have shown a different pattern of effects. While not available in the current study for a sufficient proportion of our sample, future work should integrate these biomarkers and examine the timing of skin-deep resilience.

The consequences of risk and resilience for the academic domain were distinct from those observed for mental and physical health. For grades in school, socioeconomic risk clearly mattered more for youth than resilience, with both low risk/high resilient and low risk/low resilient youth earning higher GPAs in middle and late adolescence compared to their high risk/high resilient peers. There are well-documented disparities in educational outcomes for low-income youth versus those from more advantaged families (Odgers, 2015), and the findings reported here suggest that high resilience is not enough to counteract the structural disadvantages of economic adversity.

Taken as a whole, the current study contributed to the rich knowledge base on risk and resilience through its attention to multiple developmental periods (middle and late adolescence) and its attention to a host of precursors and consequences of risk and resilience. Despite these strengths, some limitations must be acknowledged. The sample was predominantly low risk, and as such, the high-risk groups were relatively small, impacting our power to detect small effects. Despite this limitation, we were able to identify numerous significant antecedents and well-being outcomes tied to high-risk groups, but future research with a greater distribution across high and low risk should be implemented to replicate the findings reported herein, particularly given the divergence between the primary and supplemental analyses. Future research with larger samples of resilient youth could also provide more substantial evidence for potential variation in developmental timing. While the middle versus late adolescence models suggested unique patterns of effects in the predictors and consequences of resilience, the statistical tests comparing these developmental epochs were null. These null findings are likely due to issues of power given the unbalanced cell size across risk and resilience groups, and future research with larger samples (and thus greater representation of the less common groups) could test this assertion. Moreover, our measure of economic risk did not capture youth’s subjective experiences of economic disadvantage nor their experiences of discrimination tied to socioeconomic status, and each of these has been found to impact adolescents’ well-being (Fernandez & Benner, 2022; Mistry & Elenbaas, 2021; Odgers, 2015). As such, future work should explore how each of these markers of socioeconomic adversity in concert with resilience might impact subsequent well-being across developmental domains. Additionally, with the exception of GPA and one indicator of socioeconomic risk drawn from school records, the current study relied predominantly on adolescent self-report data. This approach raises the possibility of inflated associations due to shared method variance. Future research would benefit from incorporating parent or teacher reports to provide more robust estimates of the relations examined herein. Additionally, the current study utilized a single item to assess health, and while this measure of self-rated health is linked to morbidity and mortality (Idler & Benyamini, 1997), objective measures of health, including biomarkers and anthropometric measures capturing stress, bodily inflammation, and cardiovascular and metabolic functioning are important to include in future research, as initial evidence suggests physiologic dysregulation may be tied to risk and resilience profiles (Brody et al., 2013; Ehrlich et al., 2024). Finally, it is possible that the central relations under study might vary as a function of student gender or race/ethnicity, and future inquiry with large, diverse samples should examine potential variation in both the impact of interpersonal resources on adolescent risk and resilience as well as the impact of resilience on adolescents’ well-being.

Conclusion

Collectively, our findings underscore the necessity of considering economic risk and resilience in tandem and to examine resilience as a dynamic developmental process. While we found that family and school interpersonal resources promoted resilience in face of socioeconomic risks, the benefits appeared to be domain specific. Resilient youth maintained better health yet still faced academic disparities. Our results demonstrate that future research can benefit from a multisystem perspective (Masten et al., 2021) to better understand the trade-offs of coping in the face of socioeconomic adversity. Efforts to promote youth well-being should adopt a holistic approach that both strengthens the interpersonal networks that foster resilience and intervenes to address the socioeconomic adversities that necessitate high-effort coping during adolescence.

Data availability statement:

Data are publicly available.