Article: Silver, Ian A., and Joshua S. Long. 2026. “Is youth criminal legal system involvement a social-determinant for non-normative sleep patterns during adulthood?” Journal of Experimental Criminology. https://doi.org/10.1007/s11292-026-09782-0
1. Background (PDF)
Research has demonstrated that sleep duration (the amount of time an individual sleeps) and sleep quality (how well an individual is sleeping) are important for long-term health. Healthy sleep during adulthood could be difficult to establish, especially when individuals had childhood experiences that could disrupt sleep. For example, Adverse Childhood Experiences (ACEs) are amongst the strongest predictors of abnormal sleep patterns. Notably, criminal legal system (CLS) involvement before 18 expose youth to ACEs and, more importantly, could contribute to abnormal adult sleep. Prior research suggested that sleep is disrupted during confinement, where youth experience delays in the time it takes to fall asleep during periods of confinement and shortened sleep duration. However, limited is known about the effects of youth CLS involvement on adulthood sleep. The current study served as a preliminary examination, assessing if CLS involvement before 18 influences self-reported sleep duration in adulthood from 2007 to 2017, and if the effects vary as individuals age.
2. Results
Respondents reported 6.72 hours of sleep on a typical weeknight across 2007–2017, slightly below the 7–9 hours recommended for young adults, with duration declining as the cohort aged. Similarities in the self-reported hours of sleep existed across youth without CLS contact before 18, youth arrested before 18, youth incarcerated in juvenile facilities before 18, and youth incarcerated in adult facilities before 18.
The results suggested that youth incarcerated in juvenile facilities before 18 experienced a .15 increase in hours of sleep, which equates to approximately 9 more minutes of sleep when compared to youth without CLS involvement before 18. Youth incarcerated in adult facilities before 18 experienced a .27 increase in hours of sleep, approximately 16 more minutes of sleep. Arrested youth did not experience more or less sleep than their counterparts without CLS involvement before 18. Findings remained the same when counts of arrests and months incarcerated were added. The year-by-year models show these effects are not stable over time.
3. Implications
The findings align with the research that suggests that adolescent CLS involvement could influence adult health (e.g., functional limitations, mental health symptoms, premature death) and, in particular, contribute to abnormal sleep patterns during adulthood.
Increased sleep should not be perceived as a net positive, as longer self-reported sleep duration may mark poor-quality sleep driven by adulthood mental health symptoms, substance use, or chronic stress. It may also reflect the broader collateral consequences of the system. For example, individuals might sleep more if they are unemployed and/or have housing instability. Moreover, increased sleep could correspond with food insecurity.
4. Data and Methods
The study analyzed the National Longitudinal Survey of Youth 1997 (NLSY97), a nationally representative birth cohort of 8,984 individuals born between 1980 and 1984, interviewed annually from 1997 to 2011 and biennially from 2013 to 2021. The analytical sample was the 8,961 respondents under 18 at the 1997 baseline, preserving temporal ordering between exposure and covariates; they were 32–37 by 2017, when the sleep questionnaire was discontinued.
Self-reported hours of sleep on a typical weeknight in 2007, 2009, 2010, 2011, 2015, and 2017 was top-coded at 15 hours and averaged for the primary model. Self-reported arrest and confinement dates were used to build four mutually exclusive groups reflecting most serious contact before 18: no contact (n = 7,356), arrested (n = 1,317), juvenile facility (n = 179), adult facility (n = 109). Twenty-four measures spanning baseline sleep and health, demographics and residence, parental indicators, and childhood experiences were included in the model to adjust for observable differences across the CLS involvement groups.
Missing data were imputed by chained equations using a random forest estimator. A gradient boosted model predicted group membership from the covariates, yielding multigroup average treatment effect inverse probability weights (ATE-IPW). Doubly robust ATE-IPW ordinary least squares models were estimated to examine the effects of legal system involvement before 18 on average sleep and yearly measures of self-reported sleep duration, with replications under alternative distributional assumptions and without the weights.
5. Conclusion
This is the first examination of whether criminal legal system involvement before 18 shapes sleep duration in adulthood. Confinement in a juvenile or adult facility predicted longer self-reported sleep from 2007 to 2017, while arrest did not, and year-specific effects varied in both magnitude and direction. Future research should reexamine this association using official measures of sleep duration and sleep quality. Sleep-focused interventions may benefit this population, but their design depends on first identifying the mechanisms behind these disparities.
Disclosure: This research brief was prepared by Claude and reviewed/edited by Ian A. Silver.
