Abstract
Life course theories suggest that fathers’ lifetime criminal legal system contact could contribute to poor parent–child outcomes via deterioration in couple relationship quality and fathers’ behavioral health. Using paired, longitudinal data from the Multi-site Family Study (N = 1,112 couples), the current study examines the influence of three dimensions of fathers’ life course legal system contact on individual and parent–child outcomes. In fitted models, accumulated system contact in adulthood predicts fathers’ later depressive symptoms and drug misuse, which in turn predict diminished father–child relationship quality (as reported by both co-parents). Fathers who were older at the time of their first arrest had poorer relationships with their children’s mothers and, in turn, poorer behavioral health and parent–child outcomes. Conditions of confinement during fathers’ most recent prison stay do not significantly predict later parent–child outcomes, net of the influence of age at first arrest and accumulated criminal legal system contact in adulthood.
Mass incarceration’s vast reach and intergenerational implications make it a central issue in the sociology of health (Boen 2020b; Pridemore 2014; Turney 2014b; Wildeman and Wang 2017). Ballooning incarceration rates in the United States have depressed population health, including life expectancy and infant mortality (Pridemore 2014; Wildeman 2012, 2016), and contributed substantially to racial disparities in child health and well-being (Wakefield and Wildeman 2014). Indeed, parental incarceration and other forms of criminal legal system contact figure as a major potential engine of health inequality in America (Lee and Wildeman 2021; Wildeman 2012; Wildeman and Lee 2021).
Mapping the consequences of state intervention in already disadvantaged homes is a complex undertaking (Johnson and Easterling 2012; Turney and Wildeman 2013; Wildeman, Wakefield, and Turney 2013). But prior research leaves little doubt that parental—and especially paternal—incarceration makes children more vulnerable to a range of health and behavioral problems (Turney 2014b; Turney, Wildeman, and Schnittker 2012; Wakefield and Wildeman 2014). The material strains that a father’s incarceration exacerbates, from economic insecurity to housing instability, likely account for some of this heightened vulnerability (Geller and Franklin 2014; Geller, Garfinkel, and Western 2011; Schwartz-Soicher, Geller, and Garfinkel 2011). But we know less about how malleable social and behavioral factors may cushion (or worsen) incarceration’s negative impacts on parenting and child well-being (Lee and Wildeman 2021).
Stress proliferation and family process theories suggest that the stress of incarceration and other criminal legal system contact proliferates across life domains and from the affected individual to partners, co-parents, and children (Arditti 2016; Turney 2014b). Studies that apply a life course perspective suggest that the stress of criminal legal system contact proliferates over time as well. The developmental timing and accumulation of criminal legal system contact can be pivotally important in shaping individual outcomes, including health, residential stability, wealth accumulation, and system avoidance (Brayne 2014; Lorvick et al. 2018; Maroto and Sykes 2021; Warner and Remster 2021). Bringing these perspectives together, the theory of linked lives suggests that over time, the health and well-being of partners and children changes in concert with that of their family members (Besemer et al. 2017; Engels et al. 2020; Grunow and Evertsson 2021). Still, quantitative research on the consequences of mass incarceration focuses heavily on effects of an incarceration event, often failing to capture how the developmental timing, accumulation, or conditions of parents’ criminal legal system contacts may shape individual and family outcomes.
Addressing these gaps, the current study uses longitudinal, couples-based survey reports from a three-year study of 1,991 incarcerated and reentering men and their families in five states. In fitted structural equation models, greater accumulated criminal legal system contact over the life course predicts fathers’ later depressive symptoms and drug misuse, which in turn predict diminished father–child relationship quality (as reported by both co-parents). In addition, fathers who were older at the time of their first arrest tend to have poorer relationships with their children’s mothers and, in turn, poorer behavioral health and parenting outcomes. Conditions of confinement during fathers’ most recent prison stay do not significantly predict their later parenting outcomes, net of the influence of age at first arrest and lifetime criminal legal system contact. Results suggest that the life course timing and accumulation of fathers’ criminal legal system contacts have distinct and important implications for their families. They highlight the role of two malleable factors, fathers’ behavioral health and the quality of their co-parenting relationships, that may shape parent–child outcomes among fathers in contact with the criminal legal system. Findings affirm the relevance of life course and linked-lives theories for understanding how mass incarceration shapes individual and parent–child outcomes.
Background
Linked Lives in the Mass Incarceration Era
Contact with the criminal legal system is a recurrent, chronic, and nearly inevitable part of living with racial and economic marginality in the United States (Desmond and Western 2018). Criminal legal system contact has come to the fore as a major driver of health and well-being, particularly with regard to stress-related illnesses such as behavioral health conditions, cardiovascular disease, and diabetes (Archibald, Parker, and Thorpe 2018; Bruns and Lee 2020; Lee and Wildeman 2013; Massoglia 2008; Massoglia and Pridemore 2015; Massoglia and Remster 2019). Furthermore, stress proliferation and linked-lives theories suggest that major and chronic stressors like criminal legal system contact are rarely contained to the person, situation, or temporal moment in which they originate. Rather, they travel readily from one life domain to others, from earlier to later stages in the life course, and from affected individuals across their family systems (Jones et al. 2018; Pearlin, Aneshensel, and Leblanc 1997; Wickrama, O’Neal, and Klopack 2020). From arrest to incarceration, the stress of criminal legal system contact does appear to proliferate from affected individuals to their children, partners, and households (Arditti 2016; Lee et al. 2014; Lee and Wildeman 2013; Poehlmann-Tynan and Arditti 2018; Turney 2014b; Turney and Sugie 2021). Paternal incarceration, in particular, can bring greater parenting stress, harsh parenting, and child maltreatment (Austin 2016; Turney 2014a; Wakefield 2015).
Poor parent–child outcomes in families affected by criminal legal system contact likely result in part from economic strain (Geller et al. 2011; Geller and Franklin 2014; Schwartz-Soicher et al. 2011). Social and behavioral mechanisms, such as fathers’ behavioral health and relationships with their children’s mothers, may also play a role; these links remain less well studied. Furthermore, although linked-lives theory suggests that consequences of criminal legal system contact for families and children will tend to emerge over time—at developmentally sensitive time points and in connection to the emerging consequences of such contact for individual parents—studies taking a life course approach to understanding the parent’s exposure to criminal legal system contact and its unfolding consequences in the family are comparatively rare.
Behavioral Health among Fathers in Contact with the Criminal Legal System
The stress of criminal legal system contact endangers behavioral health. The well-known overrepresentation of individuals with behavioral health conditions in the criminal legal system (Fazel and Danesh 2002; Fazel, Yoon, and Hayes 2017; Yi, Turney, and Wildeman 2017) makes it difficult to disentangle selection into system contact from the effects of the contact itself (Schnittker, Massoglia, and Uggen 2012; Travis, Western, and Redburn 2014). And it also means that such contact tends to be visited on individuals with trauma histories and other behavioral health vulnerabilities, which may increase the likelihood of negative behavioral health consequences (Bruns and Lee 2020; Bryant-Davis et al. 2017; Haney 2018; Kupers 1996). The stress associated with aggressive police encounters, arrest, incarceration, and reentry from prison can each promote diagnosed psychological disorders; self-reported depressive symptoms, anxiety, and traumatic stress symptoms; and other forms of psychological distress (Geller et al. 2014; Sewell and Jefferson 2016; Sugie and Turney 2017; Turney, Lee, and Comfort 2013). Exposure to the chronic stress of incarceration, particularly prolonged incarceration, increases later risk of diagnosed psychological disorders and self-reported depressive symptoms (Boen 2020a; Massoglia 2008). Men returning from prison report that the reentry process further strains their mental health, putting them at heightened risk for psychological distress (Turney et al. 2013), bipolar disorder, dysthymia, and major depression (Schnittker et al. 2012).
Consistent with the stress-proliferation and linked-lives perspectives, parents’ behavioral health can alter parent–child relationships (Bøe et al., 2014). For fathers in contact with the criminal legal system, alcohol and other drug misuse represents an especially central issue. In the context of an ongoing War on Drugs and unmet need for community-based behavioral health treatment, many individuals with substance use disorders are repeatedly channeled into criminal legal system contact and incarceration (Begun, Early, and Hodge 2016; Belenko et al. 2016; Belenko, Hiller, and Hamilton 2013). Furthermore, men appear predisposed to externalize stress through the misuse of alcohol and other drugs (Chaplin et al. 2008; Kajantie and Phillips 2006). Alcohol and other drug misuse, in turn, can weaken parental engagement and parenting quality (Arria et al. 2012).
Depression is another central issue for parents involved with the criminal legal system. Arrest and incarceration each promote depression among young adults of parenting age (Boen 2020a) and among parents, in particular (Menting, de Castro, and Matthys 2017). Menting and colleagues’ (2017) quasi-experimental study finds that recent incarceration predisposes poor mothers to depression and cognitive distortions and may promote reduced family engagement and poorer parenting behavior. Among fathers, Turney and colleagues’ (2012) find that current or recent incarceration predicts episodes of major depression and that depression episodes are linked not only to deteriorated socioeconomic status but also to changes in the status and quality of fathers’ relationships with their partners and children.
Co-parenting among Fathers in Contact with the Criminal Legal System
Criminal legal system contact damages committed intimate and co-parenting relationships (Apel 2016; Khan et al. 2011; Western and McLanahan 2000). Even a brief stay in confinement (e.g., one month in jail) tends to disrupt cohabitation and make marriage less likely (Apel 2016). Prolonged incarceration not only separates partners but also severely hinders their communication with one another, often exacerbating intrafamilial strain and conflict during incarceration and at the time of reentry from prison (Comfort et al. 2018).
As partner and co-parenting relationships deteriorate, so do fathers’ relationships with their children. Well established in the general population (Carlson, McLanahan, and Brooks-Gunn 2008), this dynamic is especially important for fathers in criminal legal system contact, whose co-parents often serve as coaches or gatekeepers for the father–child relationship (McKay et al. 2018). Urban fathers with incarceration histories are more likely to experience the dissolution of their relationships with their children’s mothers than similar, never-incarcerated fathers—which, in turn, makes them much less likely to live with their children (Geller 2013). Among residential fathers, current or recent incarceration promotes the dissolution of their partnerships with the mothers of their children as well as a reduction in mothers’ trust for them as parents. Negative changes in the co-parenting relationship are, in turn, associated with fathers’ reduced co-parenting and reduced engagement with their children (Turney and Wildeman 2013).
Life Course Perspectives on Criminal Legal System Contact
Research on the consequences of mass incarceration has centered largely on understanding the impact of imprisonment (Kirk and Wakefield 2018; Travis et al. 2014). But studies of a singular event or status, even one as significant as imprisonment, can miss the more complex, chronic disruption of individual and family life by criminal legal system contact. Sugie and Turney (2017:720; see also Fernandes 2019), examining the ramifications of such contact for behavioral health, argue that “focusing on the health consequences of incarceration only, as is common . . . fails to capture the scope of the criminal justice system and its role in perpetuating inequality.”
Extending criminological life course theory (see Laub and Sampson 2019) to understand criminal legal system contact, critical criminologists suggest that capturing the developmental timing and accumulation of such exposure is essential to understanding its human consequences and contributions to inequality (Kurlychek and Johnson 2019; Wakefield and Apel 2016). Criminal legal system contact generates a certain life course momentum such that childhood contacts profoundly shape adult outcomes (Gilman, Hill, and Hawkins 2015) and adult contacts snowball to great cumulative effect (Kurlychek and Johnson 2019). Evidence from stress research further supports the life course view: The developmental timing and accumulation of other stressful life events (besides criminal legal system contact) is definitively important in shaping health and behavioral outcomes (Foster, Hagan, and Brooks-Gunn 2008; Serido, Almeida, and Wethington 2004) and broader health inequality (Boen 2020b; Boen, Kozlowski, and Tyson 2020; Schmeer and Tarrence 2018; Simons et al. 2019).
In emerging studies that take the developmental timing and accumulation of criminal legal system contact into account, seemingly less consequential forms of system contact come into view as defining influences on health and well-being. Arrests and convictions as well as terms of incarceration all appear to have cumulative effects on individual outcomes (in addition to the singular effects of arrest and incarceration; Kurlychek and Johnson 2019; Lorvick et al. 2018; Porter and DeMarco 2019). Ethnographic work suggests that the implications of recurrent criminal legal system contact for the family differ qualitatively from those of singular events (Comfort 2016; McKay et al. 2018). Boen’s (2020a) analysis of criminal legal system contacts through early adulthood demonstrates that the occurrence and accumulation of multiple forms of criminal legal system contact predict self-reported depressive symptoms and heightened inflammation. But the developmental timing of the first incarceration (operationalized dichotomously as “under 18” and “18 and over”) and the number of incarcerations (during youth and early adulthood only) did not influence later depressive symptoms in her analysis. Among the richest quantitative studies of this kind to date, Boen’s work points at once to the fruitfulness of the life course lens and to the many empirical questions that remain.
Much also remains unknown about the implications of conditions of confinement for individual and family life (Kirk and Wakefield 2018; Turney and Conner 2019). Various aspects of prison confinement—including (but not limited to) time in solitary confinement, frequent facility transfers, and the duration of the confinement—could alter the experience in ways that shape later individual and family outcomes (Uggen 2016; Wildeman, Fitzpatrick, and Goldman 2018; Wildeman and Wang 2017). Qualitative research suggests that time in solitary confinement may be especially damaging for behavioral health (Kupers 2017). Multiple facility transfers may exacerbate the administrative and emotional stress of an incarceration in partner and co-parenting relationships (Cochran 2019; Cochran et al. 2016). And longer-duration incarceration episodes may exert a more profound effect on individual mental health (Boen 2020a; Porter and DeMarco 2019) and co-parenting and parent–child relationships (Comfort 2008; McKay et al. 2018). Effects of such conditions on behavioral health, co-parenting, and parenting remain unknown.
Finally, life course and stress-proliferation perspectives each suggest that the consequences of criminal legal system contact will emerge around major life milestones (e.g., Mears and Siennick 2016). Reentry from prison is such a milestone not only for those returning home but also for their partners and children (Begun, Hodge, and Early 2017; Berg and Huebner 2011; Massoglia and Pridemore 2015; McKay et al. 2019; Yaros et al. 2018). Understanding how a lifetime of criminal legal system contact reverberates at this fraught and fragile moment in the individual and family life course is essential to grasping mass incarceration’s complex, intergenerational effects.
Hypotheses
It remains unknown how fathers’ criminal legal system contacts over the life course—from age at first arrest to accumulated system contacts in adulthood and conditions of the most recent confinement—shape their behavioral health and co-parenting relationships and, in turn, the degree and quality of their engagement with their children. Prior theoretical and empirical work, summarized in Figure 1, suggests that differences in parent–child outcomes after fathers’ release from prison may be explained by differences in fathers’ life course criminal legal system contact and associated variation in their co-parenting relationships and behavioral health.
Generalized Model for Effects of Criminal Legal System Contact on Parenting in Multisite Family Study Data (N = 2,224).
Applying structural equation modeling (SEM) with multiple waves of longitudinal data from returning prisoners and their co-parents, the current study tests the following hypotheses:
Hypothesis 1: Fathers’ criminal legal system contact over the life course predicts their diminished family engagement and father–child relationship quality after release from prison.
Hypothesis 2: The predictive effect of criminal legal system contact on family and parenting outcomes occurs via fathers’ poorer co-parenting relationships and behavioral health.
Data and Methods
Sample
The current study used longitudinal, couples-based data from the Multi-site Family Study on Incarceration, Parenting and Partnering (“Multi-site Family Study”). To be eligible, men had to be incarcerated in a participating state prison in one of five states at the time of study enrollment and to identify as being in a committed intimate or co-parenting relationship with a different-sex partner (for further detail, see Lindquist et al. 2018). Men and their partners were interviewed at baseline and again 9 and 18 months later. In the largest two study sites, participants were also interviewed at 34 months post baseline. Each study partner was asked to list each of their children, including minor and adult children as well as biological, adopted, and other nonbiological children. From among the male participant’s children, the research team selected a minor focal child who was co-parented with the study partner (if applicable) and closest in age to eight. The timing of men’s release from prison varied widely, but most returned home sometime during the study follow-up period. Questions about family life focused on the same study partner and children throughout the longitudinal follow-up period regardless of changes in relationship status or parenting arrangements.
Response rates generally remained at or above 75% for each interview wave. Protocols were approved by the United States Office for Human Research Protections, state departments of correction, and an Institutional Review Board at RTI International. The current study, which used deidentified survey data, was determined not to constitute research with human subjects per the United States Code of Federal Regulations (45 CFR 46, 102).
Models were constructed with data from 1,112 couples in which both members of the study couple provided information on their relationship at baseline. Models relied on men’s reports for most constructs but incorporated both partners’ reports of the main outcome variable, the quality of fathers’ relationships with their children. As shown in Table 1, sample members were generally in their early 30s, had limited formal education, and were racially and ethnically diverse. Men’s contact with the criminal legal system began in their teenage years, with the average father reporting a dozen arrests and six prior incarcerations (in jail or prison) over the course of his life. Their female co-parents reported less direct contact with the criminal legal system (and were asked much less detailed questions about such contact); they had typically been arrested one to two times in their lives. Most identified as being in an intimate relationship with their study partners and reported an average of three children (of all ages).
Multi-site Family Study Analytic Sample Characteristics (N = 2,224).
Measures
Interviewers administered an in-depth survey lasting approximately 90 minutes at each wave. The survey focused on participants’ family lives before, during, and after the father’s incarceration. The most sensitive questions, including those about co-parenting relationship quality, were answered via audio computer-assisted self-interviewing.
Criminal legal system contact
This study examined three dimensions of fathers’ criminal legal system contact based on fathers’ retrospective reports, collected at the baseline interview: their age when first arrested, their cumulative lifetime exposure, and the conditions of the most recent incarceration. Age at first arrest was captured in a single, self-report variable. Lifetime criminal legal system contact was reflected in three indicator variables: lifetime number of arrests, convictions, and adult prison or jail incarcerations. Conditions of the most recent incarceration experience were also reflected in three indicators: duration, number of transfers, and days spent in solitary confinement. All variables were based on survey items that elicited respondents’ retrospective reports of the period prior to the baseline interview. Raw responses were converted to standard scores (z scores) for purposes of model estimation.
Co-parent relationship quality
The quality of the relationship between the father and his self-identified committed romantic and/or co-parenting partner at the time of the baseline interview was measured using a shortened, seven-item version of the Dyadic Adjustment Scale (Sabourin, Valois, and Lussier, 2005) and an additional survey item that asked respondents to rate their current happiness with their relationships on a scale from 1 to 10.
Depressive symptoms
Fathers’ self-reported depressive symptoms at the time of their first postrelease follow-up interview were captured using nine items from the Center for Epidemiologic Studies screening tool for probable Depression (CES-D; Radloff 1977; Van Dam and Earleywine 2011). Respondents were invited to indicate whether they experienced a given symptom all, most, some, or none of the time during their incarceration; items were reverse-coded as needed such that higher scores indicated greater symptoms of depression. The items were: “You were bothered by things that don’t usually bother you,” “You had trouble keeping your mind on what you were doing,” “You felt depressed,” “You felt that everything you did was an effort,” “You felt hopeful about the future,” “You felt fearful,” “Your sleep was restless,” “You were happy,” and “You felt lonely.”
Alcohol and drug misuse
Fathers’ alcohol misuse and drug misuse at the time of the first postrelease follow-up interview were captured in a pair of composite variables that summed responses to the CAGE (“Cut Down, Annoyed, Guilty, Eye-Opener”) questionnaire items regarding alcohol misuse and misuse of other drugs (Buchsbaum et al. 1991). The alcohol misuse composite was augmented with an item capturing heavy drinking (Bradley et al. 1998).
Parent–child outcomes
Two dimensions of parent–child outcomes after the father’s release from prison were captured: the father’s family engagement and the quality of the father’s relationship with the focal child. The father’s family engagement was indicated by three self-report variables, each of which was measured at the first postrelease follow-up interview: how often the father participated in an activity with the focal child, with the examples “eating meals, going shopping, helping with homework, or doing something fun” provided; how often the father, study partner, and focal child participated in family-oriented activities together, with the examples “recreational activities, eating meals, or going to church”; and how often the father, study partner, and focal child enjoyed spending time together. Quality of the father–child relationship was captured with three indicators combining fathers’ and co-parents’ reports: the father’s self-rating of the quality of his relationship with the focal child, the study partner’s rating of the quality of this parent–child relationship, and a composite variable representing the father’s score on a four-item parental warmth scale; for example, “How often did you communicate with [focal child] about things s/he is interested in?” and “How often did you hug or show physical affection to [focal child]?”
Analytic Approach
Structural equation models were constructed using Stata 15.1 to test the study hypotheses while adjusting for error in the measurement of complex constructs. Models controlled for the father’s adverse childhood experiences (including how many of his parents or grandparents had been arrested and his rating of parental stability in his family of origin), race (as proxy for exposure to racialized disadvantage), educational attainment, the duration of his relationship with the focal co-parent, and whether he was incarcerated for a violent crime. The influence of fathers’ preincarceration behavioral health status (depressive symptoms, alcohol misuse score, and other drug misuse score) on the baseline quality of the co-parenting relationship was also controlled. Models further controlled for child age, based on evidence that parental warmth, engagement, and parent-reported quality of the parent–child relationship covary with child age (Amato 1989; Shearer, Crouter, and McHale 2005; Yeung et al. 2001). Descriptive statistics for all key variables included in the models are shown in Table 2.
Descriptive Statistics for Model Variables in Multi-site Family Study Data (N = 2,224).
As proxy for exposure to racialized disadvantage, this variable was coded dichotomously: 1 if respondent self-identified as White and 0 if respondent reported any other racial identity.
Education was represented by a composite score for highest level of formal schooling completed such that a higher score (scale = 1–6) indicated more schooling.
Models were estimated using full information maximum likelihood procedures (Arbuckle, Marcoulides, and Schumacker 1996) to retain all available observations. Standard errors were estimated using the observed information matrix method. Technical adjustments were made for achieving convergence and fit (e.g., adjusting starting values, adding appropriate error covariances), but additional controls or other parameters were not added to improve fit. As a sensitivity analysis, partial models were rerun among sample members who reported a racial identity other than White. (The complexity of the structural and measurement models and the number of controls precluded running a full model with this subsample, nor could a partial model be computed in the White sample only.)
Chi-square estimates were used to assess absolute model fit. For the supplemental model, standardized root mean squared residuals (SRMRS) were also estimated (SRMR could not be estimated for the main model due to missing values). Unlike the chi-square test, which can be expected to reject the null hypothesis with samples of more than 200 (even when the model fits well), the SRMR is accurate across sample sizes and model sizes and robust to data nonnormality (Maydeu-Olivares, Shi, and Rosseel 2018). Root mean squared error of approximation (RMSEA), generally preferred for confirmatory modeling in large samples (Fan, Thompson, and Want 1999; Rigdon 2009), was used to assess relative fit, as were the Comparative Fit Index (CFI) and Tucker-Lewis Fit Index (TLI). Each statistic was compared to empirically based cutoffs (Hu and Bentler 1999).
Results
Three structural equation modesl were constructed to test the pathway shown in Figure 1. Fathers’ experiences were operationalized using all available waves of longitudinal data, beginning during the (retrospectively reported) preincarceration period and continuing through the incarceration period and postrelease follow-up waves. Models tested whether fathers’ criminal legal system contact, operationalized as an observed variable representing age at first arrest and two exogenous latent variables representing lifetime exposure and the conditions of the most recent confinement, predicted their postrelease parenting, including their engagement with their children and the quality of their parenting, each of which was treated as an endogenous latent variable. Each model included direct pathways to parent–child outcomes and indirect pathways via baseline co-parenting relationship quality and behavioral health symptoms at the first postrelease follow-up interview. Co-parenting relationship quality and behavioral health were operationalized using standard measures and treated as observed variables.
Measurement Model Results
Inspection of variable loadings for each latent construct revealed a reasonable measurement model, as shown in Table 3. All coefficients were .4 or higher, suggesting that indicator variables loaded well on the intended constructs.
Measurement Model Results in Multi-site Family Study Data (N = 2,224).
Overview of Structural Model Results
Three models were run to test hypothesized pathways from lifetime criminal legal system contact to parent–child outcomes via depressive symptoms (Figure 2), alcohol misuse (Figure 3), and other drug misuse (Figure 4). The variance in outcomes that was explained by the model was .896 for the model that included depressive symptoms, .853 for the model that included alcohol misuse, and .819 for the model that included drug misuse.
Structural Model Results for Depressive Symptoms Pathway in Multi-site Family Study Data (N = 2,224).
Structural Model Results for Alcohol Misuse Pathway in Multi-site Family Study Data (N = 2,224).
Structural Model Results for Drug Misuse Pathway in Multi-site Family Study Data (N = 2,224).
Looking across the three models, results support the first hypothesis: Two dimensions of lifetime criminal legal system contact (age at first arrest and accumulated criminal legal system contact in adulthood) significantly predict later parent–child outcomes. Net of the predictive effect of other dimensions of legal system contact, however, the conditions and characteristics of fathers’ most recent incarceration do not predict later parent–child outcomes in any of the three models. Models also offer partial support for the second hypothesis by affirming that predictive effects of legal system contact on parent–child outcomes occur via fathers’ behavioral health and the quality of their co-parenting relationships. The direction of all observed relationships was as expected. The magnitude of predictive effects (as indicated in beta values) were generally modest; the strongest was for the influence of accumulated criminal legal system exposure on self-reported depressive symptoms.
Structural Model Results for Depressive Symptoms Pathway
The first model tested pathways from fathers’ criminal legal system contact to parent–child outcomes via the quality of their co-parenting relationships and their depressive symptoms.
Greater lifetime criminal legal system contact predicts greater depressive symptoms after release from prison (B = 40.09, p < .01), which in turn predicts diminished quality of the parent–child relationship (B = –.05, p < .001). Greater co-parent relationship quality predicts greater family engagement (B = .12, p < .001), which in turn predicts greater quality of the parent–child relationship (B = .45, p < .001). The direction of each of the observed relationships is as expected. All other relationships are nonsignificant.
Structural Model Results for Alcohol Misuse Pathway
The second model tested pathways from fathers’ criminal legal system contact to parent–child outcomes via the quality of their co-parenting relationships and their alcohol misuse.
In this model, higher age at first arrest significantly predicts diminished quality of the co-parenting relationship (B = –.11, p < .01), which in turn predicts greater alcohol misuse (B = –.36, p < .01) and reduced family engagement (B = .11, p < .001). Reduced family engagement predicts poorer parent–child relationship quality (B = .51, p < .001). The direction of each of the observed relationships was as expected. All other relationships were nonsignificant.
Structural Model Results for Drug Misuse Pathway
The third model tested pathways from fathers’ criminal legal system contact to parent–child outcomes via the quality of their co-parenting relationships and their drug misuse.
Paralleling results from the second model, higher age at first arrest significantly predicts diminished quality of the co-parenting relationship (B = –.11, p < .01), which in turn weakly predicts greater drug misuse (B = –.02, p < .01) and reduced family engagement (B = .11, p < .001). Reduced family engagement predicts poorer parent–child relationship quality (B = .51, p < .001). Greater lifetime criminal legal system contact also predicts greater drug misuse after release from prison. As in the other two models, the direction of each of the observed relationships was as expected, and no other relationships were significant.
Absolute and relative fit statistics and coefficients of determination for the three models appear in Table 4.
Exact and Approximate Fit Statistics and R2 in Multi-site Family Study Data (N = 2,224).
Note: df = degrees of freedom; RMSEA = root mean squared error of approximation; CFI = Comparative Fit Index; TLI = Tucker-Lewis Fit Index.
RMSEA values (.034 for all three models) were well below the recommended cutoff of .060, indicating adequate fit. Approximate fit statistics were mediocre, with CFI and TLI for each model slightly below the recommended minimum of .95. As expected with a larger sample, the likelihood ratio test comparing the hypothesized model to a saturated model was statistically significant for all three models.
Sensitivity Analysis
Three additional models were run to explore whether the observed relationships were sensitive to the father’s self-identified race, as proxy for experiences of racialized disadvantage. Due to model complexity, estimating the full models with a reduced sample was computationally infeasible, so the stratified models focused on pathways from fathers’ life course criminal legal system contact to co-parenting relationship quality and behavioral health factors (a robust body of general population research in diverse samples links these factors, in turn, to parent–child outcomes). Partial models were run in the full sample and among fathers of color (defined as fathers who identified as being any race other than White; participants predominantly described themselves as Black and multiracial). Results for the two samples were generally similar. In the model for alcohol misuse and co-parenting quality, however, age at first arrest and lifetime criminal legal system contact predicted co-parenting relationship quality in the full sample and were each nonsignificant among fathers of color. In addition, the characteristics and conditions of the most recent confinement negatively predicted alcohol misuse in the full sample but were nonsignificant among fathers of color. Approximate and absolute fit statistics were strong for all three alternate models. All CFI and TLI values were above the recommended minimum, and RMSEA values for all three models were well below the recommended ceiling.
Discussion
Key Contributions
A rich literature has brought to light the powerful implications of incarceration for health, family life, and intergenerational inequality (Lee and Wildeman 2021; Wakefield and Wildeman 2014; Wildeman and Wang 2017)—but it paints an incomplete picture of the consequences of criminal legal system contact more broadly. Focusing predominantly on effects of singular incarceration events (Kirk and Wakefield 2018), this literature fails to capture the complex and often chronic ways that criminal legal system contact and its consequences unfold over time.
Applying SEM with a unique, couples-based, longitudinal data set, the current study estimated the extent to which variation in three life course dimensions of criminal legal system contact—age at first arrest, accumulated criminal legal system contact in adulthood, and the conditions of a multiyear confinement—influence individual, couple, and parent–child outcomes. In fitted models, fathers’ lifetime criminal legal system contact predicts their family engagement and parenting quality after release from prison via changes in the co-parenting relationship and fathers’ behavioral health.
A life course perspective suggests that the consequences of criminal legal system contact may depend on its developmental timing and also its accumulation over the life course. Findings from the current study affirm this expectation: Fathers’ age at first arrest predicts later parenting outcomes via the diminished quality of co-parenting relationships and increased alcohol and drug misuse, while their accumulated contacts with the criminal legal system in adulthood predict parenting outcomes via increased drug misuse and depressive symptoms. The influence of accumulated legal system contacts manifests as expected (with more accumulated contacts predicting poorer parenting outcomes); however, the influence of age at first arrest is somewhat surprising (with older age at first arrest predicting poorer parenting outcomes).
Criminal legal system involvement that begins later in the life course, these results suggest, may be more disruptive to family relationships than earlier-life arrest. In low-income communities of color that are disproportionately targeted for early-life arrest (Boen et al. 2022), qualitative research suggests that partners and co-parents of men involved with the criminal legal system may be less likely to view such involvement as a negative characteristic of the person or relationship. Viewing such experiences as inevitable and even structural may, in turn, attenuate their negative impact on the couple relationship (Comfort 2012). By contrast, when criminal legal system contact begins later in life, it may be more disruptive to the family (because family relationships are more established) and also more likely to be viewed as a result of poor individual choices than of shared, structural adversity.
Life course scholarship further suggests that consequences of criminal legal system contact may not unfold immediately or even evenly over time; rather, their emergence may be concentrated around major life milestones. For incarcerated individuals and their families, who often spend years anticipating a return from prison (and the relief it promises from the emotional and material strains of incarceration; McKay et al. 2019), reentry is such a milestone. Earlier evidence suggests that parent–child outcomes may worsen around the time of a father’s reentry (Yaros et al. 2018). Life course and linked-lives perspectives on criminal legal system contact suggest that when a father returns home from prison, the consequences of a lifetime of criminal legal system contact may also come home—to him as well as his family. Findings from the current study affirm that possibility, suggesting that a father’s return to family and community is far from a panacea for the pains of imprisonment. Indeed, it may represent a major life course inflection point when such pains are fully felt.
Limitations and Directions for Future Research
Limitations in the data source and analytic approach mean that study results are not generalizable to a formal population of inference, nor are causal conclusions supported. The Multi-site Family Study recruited and interviewed men who were serving state prison time in five U.S. states and their intimate and co-parenting partners. The resulting data on men’s criminal legal system contact and family life are perhaps the richest available, but participants do not represent the general population of Americans involved with the criminal legal system.
Furthermore, the Multi-site Family Study enrolled participants during an incarceration and used complex survey skip patterns to ensure that questions were applicable to families with diverse structures and highly varied trajectories of incarceration and release. These design features mean that all male respondents had experienced incarceration prior to baseline and that key constructs were not measured identically and repeatedly at each longitudinal wave—making the data relatively incompatible with other effective analytic methods (e.g., fixed effects modeling and inverse probability of treatment weighting) for mitigating selection bias in studies of criminal legal system contact.
SEM represents a compelling strategy for exploring multistage pathways to parent–child outcomes in these data, given that “the causal relationships in a hypothesized mediation process, the simultaneous nature of the indirect and direct effects, and the dual role the mediator plays as both a cause for the outcome and an effect of the intervention are more appropriately expressed using structural equations” compared to regression analysis (Gunzler et al. 2013:394). But study models were subject to meaningful limitations. Although estimated structural regression coefficients were consistent across models and coefficient signs and p values were generally consistent with the hypotheses, they ranged from very strong to very weak. The use of SEM limited the number of control variables that could be included (relative to the number of available observations). Most importantly, the models do not allow us to rule out the possibility that results are artifacts of unmeasured confounding, including selection into system contact, or byproducts of differential attrition or item nonresponse. Results require further investigation with different data sources and using alternate modeling techniques to address selection bias.
Although sensitivity analyses were conducted to explore whether pathways held among fathers of color, wave-specific sample sizes were insufficient for more rigorous subgroup analysis, such as investigating whether impacts of criminal legal system contact are stable by race (as in Sugie and Turney 2017) or varied by fathers’ residential status (as in Turney and Wildeman 2013; Yaros et al. 2018). Further research on potential moderators would be valuable.
The study is also affected by measurement challenges. The focal outcome, quality of the parent–child relationship, was operationalized using parents’ self-reports and is therefore subject to reporting bias. This weakness is mitigated by combining the father’s self-report with his co-parent’s assessment of his relationship with the child; however, observational studies of parent–child interaction remain the gold standard. The retrospective reporting of criminal legal system contact represents another source of imprecision and potential bias. In the current study, SEM provides empirical feedback on the measurement approach (and model statistics indicate strong measurement models), but contacts with the criminal legal system are complex and present tremendous challenges to recall and reporting (Yan and Cantor 2019).
Building a rigorous understanding of the consequences of criminal legal system contact over the life course will require developing more precise and standard measures that combine administrative and self-reported data on system contact to tap the strengths (and mitigate the respective weaknesses) of each data source. Of course, from a life course perspective, better measurement also hinges on capturing data at developmentally sensitive time points. Dramatic variation in patterns of early-life criminal legal system contact by race and ethnicity (Boen et al. 2022) underscores the substantive importance of measuring such contacts in precise and timely ways. Timing is critical for outcome measurement as well, given that the consequences of criminal legal system contact for individuals and their families tend to emerge relative to major life course milestones, such as high school graduation (Mears and Siennick 2016), cohabitation and marriage (Apel 2016), and perhaps (as suggested in the current study) reentry from prison.
Finally, the current study suggests that research on the conditions of confinement stands to benefit from a life course perspective. Early research on the impact of conditions of confinement has produced notably varied results (e.g., Edgemon and Clay-Warner 2019; Wildeman et al. 2018). Comparing findings from the current study (which showed very limited predictive effects of the duration and conditions of an incarceration that occurred, on average, in participants’ early thirties) with Boen’s (2020b) Add Health analysis (which found significant negative effects of longer-duration confinements that occurred prior to early adulthood) is illuminating: Developmental timing could be critically important in shaping how conditions and characteristics of confinement (e.g., duration, transfers, and use of solitary confinement) shape later outcomes. Indeed, the possibility that characteristics and conditions of early-life confinements may “irreparably alter” health (Boen 2020b:301) deserves ongoing investigation.
Implications for Policy and Practice
Study results highlight two malleable aspects of health and social behavior, co-parenting relationship quality and behavioral health, that could play a role in parent–child outcomes among families in heavy criminal legal system contact (see also Turney et al. 2012). Intervening on these factors may offer an opportunity to interrupt the intergenerational transmission of inequality (McEwen and McEwen 2017).
Individual and family-based services (e.g., Tadros and Finney 2018) could help to mitigate the time-release effects of chronic stress on incarcerated individuals and their families by promoting positive coping, supporting healthy co-parenting relationships, and assisting fathers in reintegrating into the family system at the time of reentry. Adequate public funding for culturally competent, nonpunitive mental health and addiction treatment is essential, as is building clinicians’ competence regarding criminal legal system contact and its implications for families (Tadros and Owens 2020). The development and testing of tailored strategies for serving individuals and families affected by criminal legal system contact, such as the Transitions Clinic model (Shavit et al. 2017), are urgently needed in the behavioral health arena. Of equal importance are robust psychoeducational supports for healthy co-parenting in the face of severe, chronic extrafamilial stressors (Pilkington et al. 2019), including criminal legal system contact and the many adversities that follow in its wake.
Findings reinforce an urgent call to reduce the reach and intensity of criminal legal system contact and build noncarceral, nonpunitive responses to harm (e.g., McLeod 2015). We can and must design public safety policies that protect behavioral health and avoid damaging co-parenting and parent–child relationships. Amid an unprecedented, historic outmigration from American prisons to homes and communities (Shannon et al. 2017), there is no better time to understand and address the life course consequences of criminal legal system contact for individual and family well-being. Scholarly and policymaker attention to this task should reflect its urgency and pivotal importance within the broader charge of ending health disparities and securing equal social and economic opportunity.
Footnotes
Acknowledgements
I am indebted to Christopher Wildeman for guidance on this work and to the Multi-site Family Study research team (particularly Anupa Bir, Christine Lindquist, Megan Comfort, Erin Kennedy, Justin Landwehr, and Stephen Tueller) and participants for the rich data that made it possible. I truly appreciate the suggestions of the anonymous peer reviewers, which strengthened the article. Any remaining mistakes are my own.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by Grant 2004270 from the National Science Foundation. The original Multi-site Family Study data collection was funded by contract HHSP2332006290YC from the Office of the Assistant Secretary for Planning and Evaluation and Office of Family Assistance in the U.S. Department of Health and Human Services.
