Objectives. To assess the association between exposure to the US criminal legal system and well-being.
Methods. We used data from the 2018 Family History of Incarceration Survey, a nationally representative cross-sectional study of family incarceration experience (n = 2815), which includes measures of participants’ own criminal legal system exposure, including police stops, arrests, and incarceration. We measured well-being across 5 domains—physical, mental, social, spiritual, and overall life evaluation—and analyzed trends in well-being by criminal legal system exposure using logistic regression.
Results. Exposure to police stops, arrests, and incarceration were each associated with lower well-being in every domain compared with those not exposed. Longer durations of incarceration and multiple incarcerations were associated with progressively lower well-being. Those who were stopped and frisked by the police had low well-being similar to that of those who had been incarcerated multiple times.
Conclusions. Any exposure to police contact or incarceration is associated with lower well-being in every domain. More involved exposure is associated with even lower well-being.
Public Health Implications. Jail diversion and broader criminal justice reform may improve population-level well-being by reducing police contact and incarceration.
The United States has a massive criminal legal system.1–7 Contact with this system ranges from police stops to incarceration in jail and prison, all of which have expanded in recent decades.1–4 The US prison population has more than tripled since the 1970s, with a system that now incarcerates almost 2.3 million individuals—the largest incarcerated population in the world.5 An additional 4.5 million individuals are supervised in the community on parole and probation,6 with a large community police force that has grown steadily since the passage of the 1994 Violent Crime Control and Law Enforcement Act.2,4,7
Incarceration has been substantively linked to negative mental health outcomes during imprisonment, and having a history of incarceration has been linked to a worsening of chronic medical conditions, substance use disorders, mental health disorders, and even preventable deaths following release.5,8,9 However, even transient exposure to the criminal legal system may have negative implications for our nation’s health and well-being.
There is a growing body of research that highlights how contact with the police or living in a highly policed neighborhood is associated with worse mental health and psychological distress. A survey of 1261 young men in New York City revealed that individuals who reported more police contact also reported more trauma and anxiety symptoms.10 Other studies have also shown similar associations between aggressive policing or use of force and poor mental health.11–13 Further studies have shown that having a criminal record, even in the absence of being incarcerated, is associated with poor health outcomes.14 For instance, those on probation have a higher age-standardized mortality than does the general population.15 One plausible reason is the collateral consequences of a criminal conviction, such as legal restrictions that limit or prohibit people with criminal records from accessing employment, housing, education, voting, and other opportunities.
Exposure to the criminal legal system likely affects broader well-being, which is a person’s holistic condition encompassing physical health as well as emotional, social, and spiritual components. Well-being is a critically important indicator of individual- and population-level social welfare, and recently developed measures of well-being based on self-reported life evaluation have been found not only to be informative as valid measures of well-being but also to be strongly associated with key indicators of population health, such as life expectancy.16
Although there is some evidence that exposure to the criminal legal system affects well-being, the relationship has not been as closely studied in national population-based studies, leaving important questions on the full range of possible law enforcement and criminal justice system interactions and their consequences for well-being.
We examined the association between one’s exposure to the criminal legal system—including police stops, arrests, and incarceration—and his or her well-being. We hypothesized that exposure to police stops, arrests, and incarceration would be associated with lower levels of well-being, with those who have been incarcerated multiple times having the lowest levels of well-being. We also hypothesized that social support and financial well-being would moderate the trajectories of recovery of well-being after incarceration.
We used data from the Family History of Incarceration Survey (FamHIS), a nationally representative cross-sectional study originally designed to measure the national prevalence of family incarceration.17 FamHIS investigators worked with the National Opinion Research Center (NORC) to recruit a baseline sample of 4041 adults. Participants completed a brief screening tool that assessed incarceration experience in the immediate family. From this baseline sample, NORC recruited 1806 respondents with immediate family incarceration experience and 1009 respondents without immediate family incarceration to participate in the full FamHIS questionnaire, which includes items on one’s own incarceration, police contact, and well-being. This set of 2815 respondents constitutes the full sample used for this study and yielded a survey response rate of 69.7%.
The FamHIS data include a set of sampling weights, WEIGHT2, that adjusts the full-survey sample of 2815 to be representative of the US household adult population. WEIGHT2 accounts for the recruitment sampling into the baseline sample of 4041, and the stratified subsampling into the final FamHIS sample on the basis of family incarceration experience. This benchmarks the full-survey sample of 2815—including the items on one’s own criminal legal system involvement—to the US household adult population. Full details on sampling and weighting methods are detailed in Appendix Section A (available as a supplement to the online version of this article at http://www.ajph.org).
The main predictor of interest is respondents’ exposure to the criminal legal system, with 3 types of exposure that capture a range in the intensity of contact: stops by the police, arrests, and incarcerations for at least 1 night. Respondents who reported being stopped by police were asked whether they were also searched or physically “frisked” as part of the police stop. Those who reported being incarcerated for at least 1 night were surveyed for additional details about their incarceration experience, including when they were last released from incarceration (< 1 year ago, 1–5 years ago, 6–10 years ago, or > 10 years ago), how many times they had been incarcerated (once or more than once), and the duration of their only or most recent incarceration spell (1 day, 2 days to 1 month, between 1 month and 1 year, 1–5 years, 6–10 years, or > 10 years). We used a dichotomous summary measure of any criminal legal system exposure to tabulate respondents who reported having experienced any police stop, arrest, or incarceration.
The outcome of interest was self-reported life evaluation, a measure of overall well-being that originated in the 100 Million Healthier Lives (100MLives) initiative.18 This broader initiative evaluated and designed the 100MLives Adult Well-Being Assessment,18 a set of reliable and validated quantitative tools included in the FamHIS questionnaire that measures well-being overall and by specific domains.19–22 Self-reported life evaluation was measured using the Cantril self-anchoring striving scale,19 which was used in the 100MLives initiative and has been used extensively in other research on national well-being in the United States and other countries.23 Participants were asked to rank their current life satisfaction and future life prospects on scales from 0 to 10, using an image of a ladder to help visualize and conceptualize the scale (Figure B, available as a supplement to the online version of this article at http://www.ajph.org). Responses of current life satisfaction greater than or equal to 7 and future life prospects greater than or equal to 8 were classified as a “thriving” life evaluation,24 the main outcome of interest in these analyses. An increase of 1 SD in the current life satisfaction score is estimated to be associated with a 1.5-year longer life expectancy.16
The measured domains of well-being were physical health, mental health, social support, spiritual well-being, and financial well-being. Physical health, mental health, and social support were self-rated on 5-point Likert scales. Spiritual well-being was measured using a 7-point Likert scale that evaluated respondents’ sense of purpose and life meaning. Financial well-being was measured using an 11-rung ladder similar to the Cantril self-anchoring scale.
Responses were categorized as “thriving,” “surviving,” or “suffering” in life evaluation and in each domain of well-being using the 100MLives scoring system (Table A, available as a supplement to the online version of this article at http://www.ajph.org). For analyses of factors shaping postincarceration life evaluation, scales of social support and financial well-being were dichotomized into “high” and “low” categories corresponding, respectively, with “thriving” (social support scale ≥ 4; financial well-being scale ≥ 7) and “not thriving” for those measures.
FamHIS included the following covariates, which were included in these analyses: respondent age, gender, race/ethnicity, education level, income, housing type, employment status, marital status, and history of drug or alcohol addiction.
The analysis of our cross-sectional data began with a comparison of unadjusted patterns of well-being and criminal legal system exposure. We first compared proportions of respondents scored as thriving, surviving, or suffering on each well-being measure by criminal legal system exposure. We used the Kruskal–Wallis test to assess trends in well-being across types of criminal legal system exposure. To explore the possibility of these associations being driven by other covariates, we used 3 nested multivariate logistic regression models to estimate adjusted associations between criminal legal system exposure and the odds of a thriving life evaluation. First, we accounted for the key sociodemographic characteristics of age, gender, race/ethnicity, and education level. Next, we adjusted for social and economic factors: employment status, housing type, marital status, and household income. Finally, we adjusted for respondents’ addiction history. We estimated this set of models for each of the 3 types of criminal legal system exposure. We determined final model specifications using tests for collinearity, using a variance inflation factor cutoff of 2.0.
We also explored the potential dose dependence of these associations with time in 2 dimensions: duration of incarceration and time since last incarceration. First, we estimated age-adjusted trends in life evaluation across categories of duration of incarceration and across time points since release from incarceration. We then stratified the trends across time points since release by dichotomized levels of financial well-being and social support.
All statistical tests were 2-sided, with an α level of 0.05. We conducted all analyses in R version 3.5.1 (R Foundation for Statistical Computing, Vienna, Austria)25 and weighted them using FamHIS-specified weights to adjust the analytic sample to the US household adult population.
About 57% of men and 31% of women had any criminal legal system exposure. Individuals with any criminal legal system exposure were more likely to be Black (P < .001), to live in lower-income households (P < .001), and to have had a history of drug or alcohol addiction (P < .001) compared with those without exposure to the criminal legal system (Table 1).
Study Demographics by Exposure to the Criminal Legal System (CLS): Family History of Incarceration Survey (FamHIS): United States, 2018
|Variable||No CLS Exposure (n = 1584), No (%)||Any CLS Exposure (n = 1229), No (%)||Overall (n = 2815), No (%)||P|
|Age, y||< .001|
|18–24||101 (10.8)||81 (12.0)||182 (11.2)|
|25–34||340 (18.1)||310 (22.0)||651 (19.6)|
|35–54||431 (29.5)||436 (34.5)||868 (31.5)|
|55–64||298 (16.5)||232 (18.3)||530 (17.2)|
|65–74||262 (15.8)||135 (10.6)||397 (13.8)|
|≥ 75||152 (9.3)||35 (2.6)||187 (6.7)|
|Women||1013 (61.9)||457 (35.9)||1472 (51.7)|
|Men||571 (38.1)||772 (64.1)||1343 (48.3)|
|Non-Hispanic Black||171 (9.8)||226 (15.1)||397 (11.9)|
|Hispanic||246 (17.6)||165 (14.2)||411 (16.2)|
|Non-Hispanic White||1046 (63.1)||717 (62.4)||1765 (62.8)|
|Non-Hispanic Native American||7 (0.5)||19 (1.1)||26 (0.7)|
|Non-Hispanic other||114 (8.9)||102 (7.2)||216 (8.3)|
|Household income, $||< .001|
|≤ 24 999||292 (18.8)||366 (30.3)||659 (23.4)|
|25 000–49 999||436 (25.9)||346 (26.9)||782 (26.2)|
|50 000–74 999||299 (18.4)||203 (15.8)||503 (17.4)|
|75 000–99 999||233 (14.9)||134 (11.0)||367 (13.3)|
|≥ 100 000||324 (22.0)||180 (16.0)||504 (19.6)|
|Single-family home||1154 (75.1)||792 (66.2)||1946 (71.6)|
|Apartment||372 (21.0)||368 (27.5)||742 (23.6)|
|Mobile home/trailer/boat/RV/van||58 (3.8)||69 (6.3)||127 (4.8)|
|Working||931 (57.7)||748 (59.1)||1680 (58.2)|
|Not working, seeking job||82 (6.6)||102 (10.0)||184 (7.9)|
|Not working, not seeking job||446 (27.5)||304 (24.1)||751 (26.2)|
|Not working, other||125 (8.2)||75 (6.8)||200 (7.6)|
|No HS diploma||81 (8.2)||107 (14.5)||188 (10.7)|
|HS graduate or GED||303 (27.8)||262 (30.0)||566 (28.7)|
|Some college||650 (26.1)||579 (30.3)||1229 (27.7)|
|Bachelor’s degree or above||550 (38.0)||281 (25.3)||832 (33.0)|
|Never married||328 (23.2)||329 (28.1)||658 (25.1)|
|Married or living with partner||933 (58.0)||604 (48.7)||1538 (54.3)|
|Widowed, divorced, or separated||323 (18.9)||296 (23.2)||619 (20.6)|
|History of addiction||< .001|
|Yes||113 (7.0)||384 (29.3)||497 (15.6)|
|No||1425 (93.0)||791 (70.7)||2216 (84.4)|
Note. GED = general equivalency diploma; HS = high school. CLS exposure includes police stops, arrests, and incarceration. Proportions are adjusted for sample weights to be nationally representative of the US household population. Two respondents from the overall FamHIS sample were missing all CLS data, leaving 2813 respondents with any CLS exposure data.
Twenty-nine percent of respondents had ever been stopped by the police and 16% had ever been stopped and frisked by the police (Table 1). In unadjusted trends (Tables 2 and 3) compared with respondents not stopped by the police, those who had ever been stopped by the police had 0.59 (95% confidence interval [CI] = 0.44, 0.80) times the odds of a thriving life evaluation, and those who were stopped and frisked by the police had 0.45 (95% CI = 0.34, 0.62) times the odds of thriving, with a dose–response association (P for trend < .001). Those who were stopped and frisked by the police had low rates of thriving similar to the rates of individuals who had been incarcerated multiple times (Table 2). The fully adjusted models reflect a slight attenuation in these associations, although they remain statistically significant (Table 3).
Trends in Life Evaluation by Criminal Legal System (CLS) Exposure: Family History of Incarceration Survey, United States, 2018
|CLS Exposure||Thriving, %||Surviving, %||Suffering, %||No.||P for trend|
|Missing all CLS data||2|
|Police stops||< .001|
|Stopped and frisked||48.5||48.7||2.8||440|
|Incarcerated multiple times||49.0||48.0||3.0||331|
Note. Proportions are adjusted for sample weights to be nationally representative of the US household population. Of the full Family History of Incarceration Survey sample of n = 2815, a total of 2813 respondents had any CLS exposure data available.
Adjusted Associations Between Criminal Legal System (CLS) Exposure and Life Evaluation: Family History of Incarceration Survey, United States, 2018
|CLS Exposure||Thriving, No.||Not Thriving, No.||Model 1,a OR (95% CI) or P for Trend||Model 2,b OR (95% CI) or P for Trend||Model 3,c OR (95% CI) or P for Trend||Model 4,d OR (95% CI) or P for Trend|
|Police stops||< .001||< .001||< .001||.006|
|Never stopped||1247||695||1 (Ref)||1 (Ref)||1 (Ref)||1 (Ref)|
|Stopped||209||169||0.59 (0.44, 0.80)||0.60 (0.44, 0.83)||0.63 (0.46, 0.87)||0.65 (0.46, 0.90)|
|Stopped and frisked||204||229||0.45 (0.34, 0.62)||0.49 (0.35, 0.67)||0.54 (0.39, 0.76)||0.60 (0.42, 0.86)|
|Never arrested||1204||682||1 (Ref)||1 (Ref)||1 (Ref)||1 (Ref)|
|Arrested||453||411||0.59 (0.47, 0.74)||0.68 (0.54, 0.87)||0.76 (0.59, 0.97)||0.81 (0.62, 1.07)|
|Never incarcerated||1333||785||1 (Ref)||1 (Ref)||1 (Ref)||1 (Ref)|
|Incarcerated once||164||140||0.67 (0.48, 0.93)||0.77 (0.56, 1.07)||0.84 (0.6, 1.18)||0.85 (0.6, 1.21)|
|Incarcerated multiple times||160||167||0.5 (0.37, 0.69)||0.61 (0.44, 0.89)||0.73 (0.50, 1.07)||0.84 (0.56, 1.26)|
Note. CI = confidence interval; OR = odds ratio. Regressions are adjusted for sample weights to be nationally representative of the US household population. Employment status and age were collinear, so we excluded employment status from these models.
bAdjusted for age + gender + race + education level.
cAdjusted for model 2 + household income + home type + marital status.
dAjusted for model 3 + history of addiction.
Thirty-one percent of respondents had ever been arrested, which is a more intense type of criminal legal system exposure than are police stops. In unadjusted trends (Tables 2 and 3), individuals who had been arrested had 0.59 (95% CI = 0.47, 0.74) times the odds of thriving compared with those with no prior arrests. This association was somewhat attenuated in magnitude with covariate adjustment but remained statistically significant with the inclusion of social, demographic, and economic characteristics in the multivariate models; however, it was no longer statistically significant after adjusting for respondents’ history of addiction (Table 3).
Twenty-three percent of individuals had ever been incarcerated for at least 1 night. In unadjusted trends (Tables 2 and 3), history of a single incarceration (odds ratio [OR] = 0.67; 95% CI = 0.48, 0.93) or multiple incarcerations (OR = 0.50; 95% CI = 0.37, 0.69) were each associated with a dose-dependent lower odds of thriving (P for trend < .001) compared with those without incarceration experience, an association that remained statistically significant after adjusting for demographic characteristics. However, these associations were no longer statistically significant with the addition to the model of economic and social contextual factors or history of addiction (Table 3).
Longer incarceration spells were associated with roughly progressively lower proportions of age-adjusted thriving life evaluation (Figure 1a) and well-being across all domains (Figure A, available as a supplement to the online version of this article at http://www.ajph.org). However, those with the longest incarceration spells (> 1 year) were more likely to be thriving than were those with the second longest duration of incarceration (1 month to 1 year) on all measures.
Greater time since release was associated with progressively higher proportions of age-adjusted thriving life evaluation (Figure 1b). When stratified by levels of social support (Figure 1b), respondents sampled less than 1 year since release had similar proportions of age-adjusted thriving, regardless of level of social support. However, in comparisons across groups sampled temporally further from their last incarceration, those with high social support were progressively more likely to be thriving, whereas proportions thriving among those with low social support remained statistically stable. For respondents who had been released for more than 10 years, 85% were thriving among those with high levels of social support, but only 19% among those with low levels of social support. When stratified by financial well-being (Figure 1b), there were persistent differences in probabilities of age-adjusted thriving between those with high versus low levels of financial well-being, but there was no statistically significant trend across categories of time since release within the same strata of financial well-being.
In addition to the life evaluation measure of overall well-being, criminal legal system exposure was associated with a progressively lower proportion of thriving in every domain of well-being (Table Ba–c). Physical health and social well-being were especially low among those with exposures to the system. In sensitivity analyses, the progressive drop across life evaluation and each domain with exposures to police stops or arrests persisted after selecting for individuals with no incarceration history.
In the first, to our knowledge, nationally representative study of its kind, we found that each of the 3 types of criminal legal system exposure is associated with lower proportions of thriving in overall life evaluation and in every domain of well-being. There is some evidence of dose-dependent well-being associations with variation in criminal legal system exposure intensity, for example, in associations with police stops with and without searches or with single versus multiple incarcerations. Taken together, these findings provide additional evidence supporting the negative associations between one’s exposure to the criminal legal system and a holistic measure of well-being.
Contrary to our initial hypothesis, the negative association between exposure to police stops with searches and odds of a thriving life evaluation was similar in magnitude to the association estimated for those who experienced multiple incarcerations, illustrating the extent to which even lower-level contact with the criminal legal system is negatively associated with quality of life. These associations between police contact and well-being persisted in our sensitivity analyses that excluded formerly incarcerated individuals, suggesting that this association is driven by factors independent of incarceration.
Our results highlight the continued need for improved understanding of other types of criminal legal system exposure—such as police stops—which may be less severe but potentially harmful to health.10–13,26–28 There are more than 2.5 million street stops by the police each year in the United States, with about 9% involving searches and 3% involving use of force, even though at least 85% of stops do not result in either a ticket or an arrest.28 Aggressive policing practices such as stop and frisk are associated with worse health outcomes, with increased risks of exposure to physical, psychological, and sexual violence,26 and are associated with higher levels of anxiety and trauma.10 Our study is the first, to our knowledge, to show associations with a more holistic measure of well-being that includes physical health. Future studies can better characterize how exposure to police stops is associated with decreased well-being and identify potential mechanisms that promote the recovery of well-being, especially in overpoliced communities.
Finally, our findings underscore the importance of financial well-being and social support as important factors that are likely important in the recovery of well-being after incarceration. Our multivariate analyses show that the association between prior incarceration and well-being is attenuated after controlling for economic and social factors such as household income, marital status, and addiction history. Although our cross-sectional data cannot disentangle the temporality of the interplay between addiction, incarceration, and well-being, when seen together with our analyses of trends in well-being across time points among formerly incarcerated individuals, our data suggest that broader social and financial factors may be important mediators or modifiers of this association. This is consistent with previous studies on the role of social support for postrelease mental health and the role of financial security in facilitating successful reentry.29 Future studies can better characterize the role of addiction in the relationship between incarceration and well-being and can explore interventions that improve social support and financial well-being among formerly incarcerated individuals.
These findings are primarily limited by the self-reported and cross-sectional nature of the data. FamHIS study measures are vulnerable to recall bias and social desirability bias, which are challenges faced by many key data sources on incarceration and its relation to health.30 Additionally, because this was a cross-sectional study, the findings cannot address the temporality of criminal legal system exposures and well-being, much less causal effects.
Furthermore, community-level spatial factors are likely important drivers of well-being and were not included in the FamHIS. Finally, although the FamHIS draws on the nationally representative NORC panel, which allows inference to the broad population of all US noninstitutionalized adults, this address-based panel excludes individuals who were homeless or institutionalized at the time of data collection. Although the lack of currently incarcerated individuals in the study sample should not affect inferences about formerly incarcerated individuals, the lack of individuals experiencing homelessness or otherwise institutionalized individuals in the FamHIS may skew these data. This is a shared challenge of research on the consequences of criminal legal system exposure, as no nationally representative data capturing well-being and including these groups are currently available.31
Some limitations of these analyses point to potentially important avenues for future research. The FamHIS data do not allow distinguishing jail and prison contexts and also do not allow longitudinal observation over time with respect to duration or frequency of incarceration. Future exploration of variation in well-being across incarceration contexts and longitudinally over time is therefore important, especially for elucidating the role of addiction in our observed associations.
Nonetheless, our study design is strengthened by our use of a large, nationally representative study sample with high-quality sampling methods and low levels of missing data to ensure representative distributions of age, gender, race/ethnicity, and income. Our findings are compelling with their strong associations, dose gradients with degree of criminal legal system exposure, and consistency across all measures of well-being. Lastly, we used a robust measure of well-being to provide novel insight into the effects of the criminal legal system.
These analyses point to some key implications for public health and policy reform. First, this study corroborates the previously documented role of incarceration as a strong social and structural determinant of well-being in a nationally representative sample, further highlighting the importance of interventions that prevent incarceration. Second, our findings suggest the importance of social support and financial well-being in promoting well-being and the need for policy reforms that support the social, financial, and health outcomes of this vulnerable population.8,32,33 Finally, our findings on the strong relationship between lower-level police contact and well-being highlight the need for more research on the individual- and community- level effects of police contact on health and well-being. Empirically measuring well-being and lived experience can provide novel insights for health policy decisions and criminal justice reform efforts, with an aim to fostering thriving in every domain of life.
The FamHIS was funded by FWD.us. R. Sundaresh was supported for this study by the Moser Research Fellowship Award.
Note. The opinions expressed here are those of the authors and do not necessarily reflect or endorse the views of FWD.us.
CONFLICTS OF INTEREST
C. Riley and B. Roy receive funding from the Institute for Health Care Improvement and from Heluna Health to support their efforts in developing and implementing the measurement framework for the 100 Million Healthier Lives initiative. The measure of well-being used in this study was originally developed for 100 Million Healthier Lives. The remaining authors have no conflicts of interest to report.
HUMAN PARTICIPANT PROTECTION
The Yale School of Medicine institutional review board classified this study as exempt from further review because we used preexisting de-identified data.