Ethics statement

This mixed methods study was approved by the UC Davis institutional review board (IRB). Participants provided verbal consent and permission to use their deidentified quotes for research purposes. We followed the Standards for Reporting Qualitative Research (SRQR) reporting guidelines [21].

We developed our interview questions based on existing literature on unhoused youth as well as our team’s experiences working at the clinic. From January 2022 through August 2023 we invited TAY checking into the clinic or assessing other services at the youth center to participate. Participation in the interview did not affect their ability to receive health care. All participants provided informed verbal consent and were assigned unique de-identified codes to protect confidentiality. Each participant completed a series of validated instruments including the original ACEs questionnaire, a five-question Philadelphia Expanded ACEs Scale, a PHQ-9 if none was completed in the past six months, and the Brief Resilience Scale (Figures A-D in S1 Text). The Philadelphia Expanded ACEs scale builds upon the original by incorporating additional items related to neighborhood adversity and exposure to violence. The Brief Resilience Scale is a six-item tool that has been validated in over 140 studies, showing strong positive correlations with protective mental health factors and negative correlations with adverse outcomes [22].

Semi-structured interviews were conducted in a private setting to ensure participant comfort and confidentiality. Each interview included demographic questions such as gender and race/ethnicity identity. The interview guide was piloted with three TAY participants and feedback was used to revise questions for improved clarity and relevance, while still allowing space for new topics to emerge organically. With additional consent, interviews were audio-recorded for transcription and analysis.

Participants were included in the study if they had experienced homelessness at any time in their lives. Exclusion criteria included those who declined consent, did not speak English, did not complete all elements of the survey, or determination by the clinician that the individual was too ill or cognitively impaired to engage in a meaningful conversation.

Qualitative analysis

We employed a thematic analysis approach to analyze the qualitative data. The analysis process followed these established steps:

1. Data familiarization: Researchers immersed themselves in the data by reading and re-reading interview transcripts.
2. Initial coding: Two researchers independently coded the transcripts, identifying key concepts and patterns in the data.
3. Theme development: Codes were grouped into potential themes, which were then reviewed and refined through team discussions.
4. Theme refinement: Themes were checked against the coded extracts and the entire dataset to ensure they accurately represented the data.
5. Theme definition: Clear definitions and labels were generated for each theme.
6. Report production: A final analysis was conducted, selecting compelling extract examples and relating the analysis back to the research question and literature.

The analysis was an iterative process, with the research team meeting regularly to discuss and refine the emerging themes until consensus was reached on the final thematic framework. We used Microsoft Excel to facilitate data management and coding.

Quantitative analysis

Data was analyzed using SAS, version 9.4. Descriptive statistics including means, medians, standard deviations, frequencies, and percentages were generated for demographic variables and other measures among the sample as a whole. A two-level categorical variable was created to indicate TAY with high vs. low levels of reported ACEs. The median among this sample was used to classify TAY into each of the two levels. An analogous approach was taken to create a two-level categorical variable for high vs. low resilience. Resilience scores above our mean of 17 were classified as “high”, and scores 17 and below were classified as “low”. Demographic and housing characteristics were compared across participants with high vs. low ACEs and across participants with high vs. low resilience using Pearson Chi-square and Fisher’s Exact tests. Median age, resilience, ACEs and depression scores were compared across these classifications as well using nonparametric Mann-Whitney U Tests appropriate for small samples.

Quantitative results

Thirty three TAY were interviewed; five were excluded because they did not meet inclusion criteria. Of the 28 TAY included in the analysis, 36% identified as male and 18% identified as female (Table 1). In regard to gender identity, 46% identified as LGBTQ, with most LGBTQ youth being transgender or non-binary. The top three categories for self-identified racial or ethnic identity were white (42%), mixed race (27%), and black (15%). Thirty-six percent of our youth reported being in foster care at some point in their lives. At the time of the interview, 50% of participants were staying at a shelter. Twenty-one percent lived on the streets or reported they were “in between” places to stay, for example, having been kicked out of a shelter the previous night or moving from one house to another. Eighteen percent were staying in transitional living facilities (TLFs) which allow residents longer stays while they are provided resources for career development, life skills, and transition to affordable housing.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive statistics for the sample of N = 28 transitional-aged youth (TAY).

https://doi.org/10.1371/journal.pmen.0000392.t001

ACEs scores in our sample ranged from 6 to 15 with a mean of 9.7 (standard deviation [SD] = 3.0). The top ACES were parental divorce/separation (experienced by 86%), verbal abuse (82%), substance use by someone in the household, (82%), feeling unloved or unspecial (78%), and physical abuse (64%). ACE endorsement varied by gender identity; for example, among female youth, verbal abuse was ranked first whereas among male youth it was ranked fourth. Among transgender youth, feeling unloved or unspecial was ranked first.

Resilience scores in our sample ranged from 9 to 30 with a mean of 17.4 (SD = 5.2). PHQ-9 scores ranged from 1 to 27 with a mean of 14.6 (SD = 7.0). Seventy-four percent of TAY reported PHQ-9 scores of 10 or higher indicating moderate-to-severe symptoms of depression. Table 2 displays differences with regards to ACES classification. PHQ-9 scores were significantly higher among TAY in the high ACES group (p = 0.0003). Resilience scores were higher in the high ACES group (median = 18.0) relative to the low ACES group (median = 16.0), although these differences were not statistically significant (p = 0.2110). Table 3 displays differences with regards to Resilience classification.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Differences in demographic and other characteristics between Low and High ACES classifications.

https://doi.org/10.1371/journal.pmen.0000392.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Differences in demographic and other characteristics between Low and High Resilience classifications.

https://doi.org/10.1371/journal.pmen.0000392.t003

Qualitative results

In the following sections, direct quotes from our participants explore three prominent themes of trauma and three themes of what we perceived as “resilience-promotive” factors that interviewees considered positive drivers in their life.

Trauma profiles

Family violence and abuse emerged as the central thread connecting the trauma experiences across our sample. These events systematically dismantled family units that were often already vulnerable and poorly functioning. When biological families dissolved, participants faced cascading consequences: fractured relationships, housing instability, threats to personal safety, educational disruptions, and severe financial stress.

The pathway to homelessness for these youth typically involved being shuffled between multiple living arrangements. Some were forcibly removed from one location and compelled to find another, while others chose to leave when their current housing became unsafe or unstable. Having grown up without consistent support and nurturing, these youth entered young adulthood poorly equipped to navigate the complex web of personal, interpersonal, and structural challenges they encountered.

Participants who had experienced foster care or group homes reported the greatest instability and weakest support networks. They frequently described how seemingly minor rule violations—arriving ten minutes past curfew or smoking outside a shelter—resulted in immediate termination of their housing arrangements. This punitive approach left many with nowhere to turn, accelerating their path to street homelessness. This echoes concerning findings from Los Angeles cohort studies, where significant numbers of transitional-age youth identified punitive group home policies as direct catalysts for their transition to street homelessness [23]. This pattern stands in stark contrast to European approaches that prioritize harm reduction over rigid rule enforcement in youth shelter systems, suggesting alternative models that might better serve this vulnerable population [24].

Our findings align with broader patterns documented in homeless populations. The prevalence of family violence and abuse in our sample mirrors research on adult homelessness, where 62% report childhood adversity as a precursor to housing instability [25]. However, our participants’ experiences reveal a key distinction from studies of homeless families, where parental bonds typically endure as protective factors even amid economic hardship [26]. For our participants, these fundamental family connections had been severed entirely.

Resilience challenges

Understanding resilience in unhoused youth proves complex because resilience emerges from both innate characteristics, or assets, and environmental factors, or resources, in accordance with resilience theory [18]. The Brief Resilience Scale (BRS) measures resilience in terms of assets—the ability to bounce back from adversity—but this approach captures only a snapshot of an individual’s stress recovery capacity. It fails to provide a comprehensive picture of how people overcome hardship across multiple life domains, build healthy relationships, and pursue meaningful goals.

This limitation became apparent in our findings. Even participants who scored in the low resilience group on the BRS demonstrated remarkable ability and determination to employ diverse coping mechanisms and displayed various adaptive characteristics, as seen in Table 4. This observation aligns with what researchers call “Windle’s paradox”, the reality – seen in recent adult homeless studies -- that homelessness itself often promotes resilient behaviors and complex adaptive strategies regardless of how individuals score on psychometric measures [27,28].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Evidence-based models for promoting resilience in unhoused youth populations.

https://doi.org/10.1371/journal.pmen.0000392.t004

An intriguing pattern emerged when we examined housing status alongside resilience scores. Individuals in transitional living programs (TLPs) reported slightly higher resilience scores than those in more precarious housing situations. These data and our anecdotal observations inform our belief that stable housing creates a foundation that allows people to feel safer and more supported, knowing their basic needs are temporarily secured. This security, in turn, enables them to feel more hopeful about the future and pursue other goals such as employment, friendships, intimacy, and personal confidence—a progression that aligns with Abraham Maslow’s hierarchy of needs [29].

However, the relationship between resilience and housing may be bidirectional. While stable housing may foster resilience, it’s equally plausible that individuals with higher resilience are better positioned to advocate for themselves and successfully access more permanent housing arrangements. This creates a complex dynamic where resilience both influences and is influenced by housing stability.

Housing stability as a resilience catalyst

The relationship we observed between transitional living programs (TLPs) and higher resilience scores aligns with a growing body of evidence demonstrating that housing-first interventions enhance coping capacity across diverse populations. This connection extends beyond our specific sample—recent studies have documented similar patterns, showing that housed transitional-age youth maintain significantly better odds of sustaining employment compared to their street-dependent peers [24]. These findings suggest that stable housing creates a foundation that enables young people to engage more successfully with other life domains.

Our interpretation of these findings through Maslow’s hierarchy of needs offers one lens for understanding this phenomenon. However, this hierarchical model may not capture the full complexity of recovery pathways for homeless youth. Some researchers challenge the assumption that basic needs must be fully met before social and emotional development can occur, instead arguing that social connections often serve as the primary catalyst for positive change in this population [26].

This alternative perspective suggests that meaningful relationships and social support may be equally or more important than housing stability in fostering resilience. Rather than viewing needs as strictly hierarchical, this model proposes that housing, social connection, and personal development can occur simultaneously and reinforce each other in complex ways. Resource development for homeless youth should therefore adopt as multipronged an approach as possible, fostering support and programming in all these domains. In practice, we have observed homeless youth pursuing relationships, employment, and housing opportunities in tandem, with progress in any one area supporting advancement in others.

Implications for policy and practice

While trauma serves as a universal pathway to homelessness, resilience strategies vary dramatically by age and context. Our participants predominantly engaged in “survival-driven resilience”—resource-sharing, mutual protection, and collaborative problem-solving with peers. This contrasts sharply with older adult populations who more frequently utilize formal community services, reflecting both developmental differences and systemic barriers that make formal services less accessible to youth [25].

These findings reveal a critical gap in service delivery: most interventions emphasize individual case management designed for adults, yet our participants tended to respond better to peer-based survival networks. Effective interventions for transitional-age youth must blend housing stability with trauma-informed peer support models, honoring the collaborative strategies youth have developed while providing stable foundations [24]. Furthermore, in accordance with Windle’s paradox we should recognize that these youth have started to develop a set of self-advocacy and learned resilience skills unique to this population, and programs should seek to nurture, reinforce, and introduce these skills in a safe setting.

Limitations

Several limitations must be considered when interpreting our findings. This mixed methods, cross-sectional study drew from a convenience sample of youth presenting to a single free clinic serving unhoused transitional-age youth. Most significantly, our data potentially excludes a substantial population of unhoused young adults who avoid formal services entirely. This population may differ systematically from our participants in important ways, including trauma exposure, coping strategies, trust in institutions, or severity of mental health challenges. Their exclusion limits our understanding of the full spectrum of experiences among unhoused youth.

The small sample size and single geographic location further constrain our findings. These factors reduced the statistical power of our analyses and limit generalizability to unhoused youth in other regions where service systems, local policies, or demographic characteristics may differ substantially. Consequently, our findings should be viewed as hypothesis-generating rather than definitive, and the associations we identified do not establish causality.

Methodological constraints also shaped our data collection. The semi-structured interview format, while allowing for rich qualitative insights, meant that participants responded to open-ended questions in highly varied ways. The clinical setting imposed time limitations that may have prevented natural discussion of certain topics—for instance, a participant’s failure to mention current hobbies or interests may reflect conversational flow rather than the absence of these activities. Additionally, to meet journal word limitations, we necessarily truncated or selected portions of longer participant quotes, which may not fully capture the complexity of their experiences.

These limitations underscore the need for future research employing larger, multi-site samples and longitudinal designs that can better establish causal relationships and include harder-to-reach populations of unhoused youth who remain outside formal service systems.

Conclusion

Our findings highlight a critical need to move beyond simply addressing trauma to actively cultivating positive experiences that can counteract its effects. Future research should explore the potential of “counter-ACES”—positive childhood experiences such as stable family relationships and high emotional support—to guide intervention development [11,30]}. These advantageous experiences have been linked to improved adult outcomes including higher productivity, stronger internal locus of control, enhanced adaptive traits, and better cardiovascular health.

Residential stability represents one of the most powerful counter-ACES interventions available. Housing First models, which prioritize safe housing for anyone experiencing homelessness—particularly those with mental health or substance use challenges—have demonstrated significant benefits [31]. These include reduced mental health symptoms, decreased criminal justice involvement, lower emergency department utilization, and improved self-reported quality of life. Our findings support this model but suggest that housing alone is insufficient; effective interventions must address the complex interplay between trauma, resilience, and peer support that characterizes this population. In Table 4, we present intervention models grounded in resilience theory that integrate trauma-informed peer support—an approach that addresses both the immediate and long-term needs of this vulnerable population.

Public health approaches to youth homelessness must involve primary prevention to target the social determinants that create pathways to homelessness such as family violence, systemic inequities, and institutional failures that impact at-risk youth. Secondary prevention efforts should focus on building resilience among youth already experiencing housing instability, recognizing that traditional adult-focused services may not adequately serve transitional-age populations.

While we cannot undo the adverse experiences our participants have endured, transitional-age youth centers can implement trauma-informed, evidence-based strategies that nurture resilience-building behaviors. The path forward requires recognizing that unhoused youth possess inherent strengths and adaptive capacities that, when properly supported, can serve as foundations for recovery. By building interventions that honor these existing resilience strategies, we can create more effective response systems for this unique population.