Formerly Incarcerated Parents and Their Children

Abstract

The negative effects of incarceration on child well-being are often linked to the economic insecurity of formerly incarcerated parents. Researchers caution, however, that the effects of parental incarceration may be small in the presence of multiple-partner fertility and other family complexity. Despite these claims, few studies have directly observed either economic insecurity or the full extent of family complexity. We study parent-child relationships with a unique data set that includes detailed information about economic insecurity and family complexity among parents just released from prison. We find that stable private housing, more than income, is associated with close and regular contact between parents and children. Formerly incarcerated parents see their children less regularly in contexts of multiple-partner fertility and in the absence of supportive family relationships. Significant housing and family effects are estimated even after we control for drug use and crime, which are themselves negatively related to parental contact. The findings point to the constraints of material insecurity and the complexity of family relationships on the contact between formerly incarcerated parents and their children.

Introduction

In the context of historically high incarceration rates in the United States, parental incarceration has become highly prevalent for poor children, particularly for African Americans whose parents have little schooling. Sykes and Pettit (2014) estimated that 62 % of black children whose parents have not completed high school will experience the imprisonment of a parent by age 17 (see also Wildeman 2009). The great prevalence of parental imprisonment motivated research on the effects of incarceration on child well-being. Parental incarceration was found to be associated with reduced family income for children, an increased risk of child homelessness, aggressive behavior, depressive symptoms, and diminished school achievement (for reviews, see Foster and Hagan 2015; Johnson and Easterling 2012; Travis et al. 2014).

Lying behind research on parental incarceration are assumptions about the material wellbeing of formerly incarcerated parents and the complexity of their family relationships. Researchers often appeal to the economic insecurity of incarcerated fathers to explain the negative effects of parental incarceration. Diminished earnings and employment after prison may undermine parental support (e.g., Geller et al. 2011; Turney and Wildeman 2013; Wakefield and Wildeman 2013). Family complexity, on the other hand, may be a source of heterogeneity in incarceration effects. Incarceration effects may be small where parents are nonresident, are less engaged with their children before incarceration, or provide for the children of multiple partners. Indeed, some studies have found that the negative effects of incarceration on children are largest in stable, well-resourced families with resident parents who were positively engaged prior to imprisonment (Turney 2016; Wakefield et al. 2016; Wildeman and Turney 2014).

Although researchers have claimed that economic insecurity and family complexity shape the relationships between parents and children after incarceration, little direct evidence exists. Most research is based on child-centered data designs in which the life conditions of noncustodial parents are often unobserved. The network of family relationships in the presence of multiple partners is also typically unobserved and unanalyzed.

We contribute to research on parental incarceration by studying parental coresidence and other contact with children at a vital moment of transition: in the year after imprisonment. Our analysis uses a unique data source, the Boston Reentry Study (BRS), which records the material life conditions of formerly incarcerated parents and all their children. The data design extends earlier research in two ways. Analysis includes detailed information on the socioeconomic status of formerly incarcerated parents, regardless of whether they are living with their children. We also observe the constraining influence of multiple-partner fertility and the status of relationships with multiple partners. Our quantitative analysis is supplemented by qualitative interviews that help describe the content and sources of parental contact. By using a specialized sample in which all parents have been recently released from incarceration, the analysis helps explain heterogeneity in parent-children relationships in the year after parental imprisonment.

The analysis yields three main findings. First, regular contact between formerly incarcerated parents and their children depends on parents’ stable housing. Unstably housed parents are less likely to be in regular contact with their children, and stable housing—more than income—is related to positive parent-child relationships in the first year after incarceration. Second, complex and unsupportive family relationships are associated with reduced contact with children. Children living in different households clearly limit the possibility of parental coresidence. Qualitative data show that other family members—particularly children’s grandmothers—play a key role in managing relationships between incarcerated parents and their partners and children. Third, family relationships face the ongoing stress of drug use and crime. Hard to capture in a regression model but clearer from qualitative interviews, housing stability, congenial relationships among parents, and desistance from crime and drug use often operate together to promote contact between formerly incarcerated parents and their children.

Contact Between Children and Formerly Incarcerated Parents

Research on the effects of parental incarceration has studied samples of children, comparing those whose parents have and have not been incarcerated. Studies have relied on data from the Fragile Families and Child Well-Being Survey, the National Longitudinal Study of Adolescent to Adult Health, and the Project on Human Development in Chicago Neighborhoods (e.g., Foster and Hagan 2009; Geller et al. 2012; Turney 2016; Wakefield 2015; Wakefield and Wildeman 2013). Child samples have contributed significantly to our understanding of the effects of parental incarceration, but the data design has two limitations. First, when parents are nonresident and not interviewed, little information is available about their life conditions after incarceration, even though those conditions may influence their support of children. Second, complex family relationships—involving multiple-partner fertility and nonresident parenthood—may be important sources of variation in the relationship between children and parents, but analysis of a focal child excludes other parent-child relationships from analysis (Sykes and Pettit 2014). For example, a father with two children, each with a different mother, may be highly involved with one child but uninvolved with another (Tach et al. 2014). The structural context of multiple-partner fertility is often unobserved with a child-centered design. As a result, a low level of parental contact might be misunderstood as a low level of parental commitment.

Limitations of data design are addressed by the BRS, which followed a cohort of men and women for one year after release from prison in Massachusetts to the Boston area (Western et al. 2015). The BRS sample is representative of prison releasees in Massachusetts. The sample is small (N = 122) but includes respondents whose socioeconomic and demographic characteristics resemble those of prison releasees across the United States. BRS respondents were asked at the baseline survey, one week before prison release, to list all their children. To measure social parenthood, respondents were also asked to name all other children to whom they felt like a father figure or a mother figure. Those named as social children were mostly the sons and daughters of relatives and partners. At each wave of the survey, respondents were asked about contact with all biological and social children.

Table 1 reports descriptive statistics on the parental status of all the BRS respondents. In a sample of 122 men and women just released from prison, 78 % reported they had at least one biological child or one social child. Two-thirds of all respondents reported biological children, naming an average of 2.1 children. The children had a mean age of 14 years (median age of 13), slightly older than the children studied in earlier research on parental incarceration.¹

Table 1.

Mean parental status of formerly incarcerated men and women: Boston Reentry Study

Relationship to Child	Percentage With Children (1)	Mean Number of Children (2)	Mean Children’s Age (3)	Number of Parental Partners (4)
Biological	66.4	2.1	14.7	1.9
Partner’s	9.0	1.2	17.2	1.5
Relative’s	34.4	2.0	14.0	1.4
Social	41.8	1.9	14.4	1.5
All Relationships	77.9	2.8	14.6	2.2
Sample Size	122	95	270	70

Notes: Samples sizes for each column are based on (1) all respondents, (2) respondents with children, (3) all reported children, and (4) all respondents with two or more children. Biological includes all the respondents’ own children reported at the baseline interview. Social includes all children to whom respondents felt like a “father/mother figure” at the baseline interview, including the children of partners and relatives.

Table 1 also shows the average number of parental partners for formerly incarcerated parents with two or more children. Respondents with two or more biological children reported an average of 1.9 parental partners, and 68 % had children with at least two parental partners, a rate of multiple-partner fertility significantly higher than the estimated rates in other studies.² In short, children of BRS respondents are somewhat older than those studied in earlier research, and multiple-partner fertility is modal among parents with at least two children.

Patterns of contact and coresidence in the year after prison are reported in Fig. 1. These measures form the dependent variable for our study. The figure shows data at six time points. The baseline interview in prison recorded contact with children during incarceration—including phone calls, letters, and visits—and coresidence prior to arrest. Follow-up interviews were conducted at one week, two months, six months, and one year after prison release. Respondents were asked whether they were living with their children or had otherwise been in contact since the last interview, and how frequently they had been in contact. Notably, approximately one-quarter of children were living with their parent, prior to arrest.

Fig. 1.

Children’s rates of contact with formerly incarcerated parents in the first year after prison release. Total number of children = 270; total biological children = 173

The top panel of Fig. 1 shows that only 10 % of all biological and social children were living with their formerly incarcerated parents or social parents in the year after prison. A similar proportion reported occasional contact, measured by any contact between survey waves. Approximately one-half of all biological and social children were in weekly contact with their formerly incarcerated parents. Few parents lived with their biological children immediately after incarceration, but 60 % to 70 % were in weekly contact.

Parent-Child Relationships After Incarceration

Recent research with the BRS on family contact in the first year after prison focused on the mothers, aunts, and older sisters of men and women just out of prison (Western et al. 2015). These older women often provided housing and financial support to recently incarcerated adult relatives. Earlier research with the BRS contrasts with a long-standing research effort to examine relationships between formerly incarcerated parents and their children. The current focus on parental contact with children extends research with the BRS by examining parenthood as a marker of social integration after incarceration.

Three main theories account for variation in parent-child relationships following parental incarceration: the economic insecurity of parents after prison release, the complexity of family relationships, and criminal involvement and drug use that accounts for selection into incarceration.

Economic Insecurity

The negative effects of parental incarceration on child well-being are often traced to the economic insecurity of formerly incarcerated fathers. Research on child support has shown that father’s financial contributions are linked to contact with children and improved well-being (Amato and Gilbreth 1999; Nepomnyaschy 2007). Economic contributions may motivate nonresident fathers to be more involved with their children, and mothers may be more receptive to financially supportive fathers. Incarceration, however, fuels unemployment and child support arrears, eroding fathers’ capacity to support their children (Geller et al. 2012; Holzer 2009; Wakefield and Wildeman 2013).

Economic insecurity following incarceration takes many forms besides the income losses associated with unemployment. For example, housing insecurity is perhaps the most urgent material hardship. Temporary and unstable housing are likely to limit parental involvement. Formerly incarcerated men and women often live in shelters or transitional housing programs, or are homeless on the street (Herbert et al. 2015; Western et al. 2015). Street homelessness and extreme residential instability are likely to undermine the routines of consistent parenting. Temporary housing and dormitory living in shelters or residential programs reduces the possibility of shared custody, overnight visits, or regular meetings with children. Seldom explored in earlier research, housing insecurity also includes reincarceration. More than 20 % of those released from state prison return to incarceration within one year, eliminating the possibility of coresidence and reducing regular contact between parents and children (Durose et al. 2014).

Immediately after incarceration, private housing is typically unaffordable, and the formerly incarcerated mostly stay with family, change residences frequently, or use temporary housing services (Herbert et al. 2015; Western et al. 2015). As a venue for family relationships, housing immediately after incarceration thus describes a distinct channel for parent-child contact that is not reducible to income. The following analysis examines how coresidence and regular contact with children are associated with income, unstable housing, and reincarceration after prison release.

Complexity of Family Relationships

Relationships between formerly incarcerated parents and their children often unfold in a complex network of family relationships. Multiple-partner fertility, biological and social parenthood, and the distinct bundle of relationships accompanying motherhood and fatherhood all index the complex structure of family relationships for parents who are sent to prison. With multiple-partner fertility, maintaining contact with children in different households depends on successfully managing relationships with several parental partners (Tach et al. 2010). Tach et al. (2014) observed that parenting is distributed unevenly across the children of two or more parental partners in a study of poor fathers. We hypothesize that levels of parental involvement—coresidence and regular contact with children—will be lower for parents with more than one parental partner.

Although social parenthood has been widely observed alongside biological parenthood in poor communities (Stack 1974), parents tend to have longer and deeper histories of support with their biological children. In our interviews, respondents’ emotional energies were more often absorbed by their biological children. Investments of time and money in the parental role were concentrated more on biological than on social children. Our analysis includes biological and social children, but we expect parents to be more involved with their biological children.

Our analysis also includes formerly incarcerated mothers as well as fathers. Maternal incarceration has been found to have larger effects on family life (Arditti 2012:59, 67; Siegel 2011; Wildeman and Turney 2014). Relatively large effects of maternal incarceration are related to the higher rates of custody and child contact for mother compared with fathers who go to prison (Kruttschnitt 2010; Kruttschnitt and Gartner 2003). We expect that mothers are more likely to be coresiding or in regular contact with their children, even in cases where they do not have formal custody.

The quality of family relationships is closely tied to their complex structure. The quality and history of relationships between parents have been linked to coresidence and regular contact between and their children after incarceration (Braman 2004; Edin and Nelson 2013; Nurse 2002). Edin and Nelson (2013) described the gatekeeping role of partners as part of the “new package deal” in which nonresident fathers—motivated by a relationship with their children—must navigate relationships with their children’s mothers. Formerly incarcerated mothers and fathers will likely be in close contact with their children when they have good relationships with their parental partners. One measure of relationship quality in our analysis is based on retrospective reports of parental support prior to incarceration. Because visits and phone calls with children during incarceration typically require the active support of the unincarcerated parent, we also study variation in contact with children before and after release. In the transitional year after prison release, relationships with partners are often highly fluid. We try to capture the ebb and flow of romantic relationships after incarceration with a measure of feelings toward one’s partner that is updated at each wave of the survey.

Criminal Involvement and Drug Use

Finally, men and women who go to prison often have histories of criminal involvement, domestic violence, and heavy drug or alcohol use. Research on the effects of parental incarceration has tried to account for this nonrandom selection into incarceration (Murray and Farington 2008 review these efforts). Recent studies have also emphasized the heterogeneity of incarcerated parents depending on drug use and involvement in crime (Siegel 2011; Wakefield et al. 2016; Wildeman and Turney 2014). We examine the heterogeneity of formerly incarcerated parents with data on prior family violence, substance abuse, and arrests. These measures provide more detailed measurement of the behavioral correlates of incarceration than in prior research.

In sum, economic insecurity, the complexity of family relationships, and crime and drug use are all likely to be closely related to parental contact after incarceration. Testing these hypotheses requires direct observation of parent-child relationships after incarceration with data from the network of kin relations that includes children from multiple partners. Such data should also measure the material conditions of life for formerly incarcerated parents, even when they are not living with their children.

Data and Methods

Child-centered sampling designs have made valuable contributions to our understanding of the effects of parental incarceration but analysis has often been incomplete. Characteristics of noncustodial parents are often unobserved, and the child-centered design has bracketed the full network of family relationships from analysis. Thus theories of economic insecurity following incarceration and the role of family complexity in parental involvement are often conjectured but rarely studied directly.

Data from a cohort of men and women newly released from prison add a new dimension to research on parental incarceration. By recording all the children (i.e., biological and social) of a respondent close to prison release and following up for one year, the BRS includes rich data on family complexity and the evolution of relationships immediately after incarceration. At each interview wave, respondents were asked about their contact with all children named at the baseline in-prison interview. Data were collected on coresidence with children and parental support after incarceration. Data are also available on the respondents’ positive or negative feelings toward parental partners. In addition to interviews with formerly incarcerated respondents, we conducted supplementary interviews with family members. Interviews were largely structured, although qualitative data are also available that help illustrate the empirical content of the quantitative measures.

The dependent variable for this analysis is a four-category measure of the parent-child relationship for all children reported at baseline. At each follow-up interview, each child was coded as (1) living with the respondent; (2) in weekly contact; (3) in occasional contact, at least once between interviews; or (4) out of contact altogether. Coding the dependent variable in this way groups self-reports of daily and weekly contact. Similar to Hissel et al. (2011), we found in supplementary interviews with family members that parents sometimes overreport contact with children. Grouping daily and weekly contact helps guard against overestimates of parental involvement. We further explore possible biases and the content of contacts between parents and children in the qualitative interview data presented later.

The four-category measure of parental contact is modeled with a multinomial logistic regression in which the level of contact is contrasted with the baseline category of no contact between parent and child. The multinomial model allows differences in the associations between covariates and different levels of parental contact. For example, respondents with several partners are unable to live with all their children but may nevertheless be in regular contact. We report results for all biological and social children and for biological children separately. Regression standard errors are adjusted for clustering at the child level.

Table 2 reports descriptive statistics on the key covariates used in the regression analysis. Our analyses are based on 259 of 270 children reported at baseline by BRS respondents for whom data are reasonably complete. The economic insecurity of formerly incarcerated parents is captured by measures of housing and income. Unstable housing is indicated by residence in a shelter, transitional housing, or a rooming house; dividing time between different residences; or homelessness on the street. In each case, housing is temporary; and in most cases, respondents are living in group quarters, not conventional households that children might easily visit. Approximately one-quarter of children have insecurely housed parents at any point in the year after prison release. Transitional housing and homeless shelters account for nearly all the observed unstable housing.

Table 2.

Means of covariates used in regression analysis, by child’s relationship to formerly incarcerated parent

	Biological Children	Social Children	All Children
Parent’s Life Conditions After Prison
Unstable housing	0.309	0.193	0.268
Reincarcerated	0.032	0.033	0.033
Monthly income ($100s)	12.263	10.683	11.703
Demographics of Incarcerated Parent
Female	0.178	0.196	0.184
Number of parental partners	2.389	2.789	2.531
Quality of Family Relationships
Contact in prison	0.829	0.789	0.815
Pre-arrest support	3.096	3.159	3.118
Feelings toward partner	3.553	3.080	3.485
Parent’s Crime and Drug Use
Using drugs/alcohol	0.448	0.661	0.523
Arrested	0.094	0.089	0.092
Prior restraining order	0.527	0.459	0.503
Number of Children	162	97	259
Number of Child-Waves	596	327	923

Notes: Number of children reported by respondents is 270. Number of parent respondents is 95. Pre-arrest support and feelings toward partner are measured on 5-point scales. Because of missing data, feelings toward partner is observed for 528 child-waves and 158 children.

Another measure of housing instability indicates parental reincarceration in the time since the previous interview. Although a return to custody is reported in only 3 % of child-wave observations, 16 % of all BRS parents had returned to custody at some point by the 12-month interview. Economic status is measured by total income from employment, government programs, support from family and friends, and illegal activities. Parents report mean monthly incomes between $1,100 and $1,200, and nearly all record poverty-level incomes at some point in their year after incarceration. Family complexity is measured by the number of parental partners reported by the respondent. Each child in the sample has a formerly incarcerated parent who has, on average, 2.5 parental partners.³ The demography of family structure is controlled by indicators for mothers and for biological children.

The quality of family relationships is measured by three variables. The first measure indicates respondents who had contact with their children—through phone calls, letters, or visits—during incarceration. The second measure of relationship quality, recorded at baseline, counts respondents’ reports of providing daily care, play and activities, emotional support, discipline, and financial assistance to form a 5-point scale for support of children prior to arrest. The third measure, recorded at each post-release interview, is a 5-point scale capturing the respondents’ feelings toward their partners, from very positive to very negative. Not measured for all social children, feelings toward partner can be included only in the analysis of biological children.

Finally, the analysis controls for the respondents’ involvement in crime, domestic violence, and drug use. Criminal involvement is measured with a dummy variable indicating whether the respondent was arrested since the last interview. Drug and alcohol use is self-reported, and more than one-half the sample reported substance use throughout the survey period. The analysis also includes a time-invariant measure of domestic violence, indicating respondents with restraining orders in their official criminal record. Approximately one-half the sample had a restraining order that restricted contact with former partners, crime victims, or family members.

In addition to these covariates, all the regression analyses control for the child’s age and race or ethnicity. Other specifications were also estimated, including controls for length of prison stay; parole or probation status; diagnoses of depression, anxiety, chronic pain, or disease; respondent’s family background; and social isolation immediately after prison release. The reported results are robust to the inclusion of these other variables. A sensitivity analysis exploring robustness of the results to alternative models is reported in the appendix. We also conducted an analysis restricted to just the fathers in the sample. The results for fathers only, similar to those presented here, are reported in Table 7 in the appendix.

Several of the hypotheses are causally motivated, but we present the following results as a description of the relationship between parents and children after incarceration. Some predictors measured before prison release are clearly causally prior. Still, relationships with partners after release are likely to be endogenous to parent-child relationships, and any causal effect is overestimated. Even without a strong causal interpretation of the results, a detailed description of parent-child contact remains informative about the well-being of families with formerly incarcerated parents.

Results

Multinomial regressions describe the log odds of contact between parent and child compared with a baseline of no contact. We also supplement the quantitative results with qualitative cases to illustrate the content and process of parent-child contact after incarceration.

Regression Results

Table 3 reports the multinomial regression results from the analysis of the full sample of biological and social children reported by the BRS respondents; Table 4 reports the results for biological children only. Among the measures of economic insecurity after incarceration, contact with children is only weakly related to monthly income but strongly related to housing. Parents who were unstably housed in shelters or transitional housing or who were homeless were very unlikely to be in regular contact or coresiding with their children.⁴ The coding of contact with children includes prison visits and phone calls, but these are highly unlikely for children whose parents whose have returned to incarceration. Housing status may depend on income, but even when housing is excluded from the model (results not shown here), income effects are mostly small and insignificant.

Table 3.

Multinomial logistic regression analysis of contact with formerly incarcerated parents for all biological and social children: Boston Reentry Study

	Occasional Contact(1)	Weekly Contact (2)	Coresidence(3)
Constant	–1.596 (1.91)	–3.329** (3.99)	–1.767 (1.40)
Life Conditions After Prison Release
Unstable housing	–0.373 (1.18)	–0.995** (3.71)	–2.203** (4.49)
Reincarcerated	–1.674* (2.34)	–2.388** (3.50)	–16.235** (22.78)
Income	–0.006 (0.48)	0.007 (0.63)	0.008 (0.54)
Demographics of Incarcerated Parent
Biological child	0.477 (1.37)	1.634** (4.98)	0.514 (1.07)
Female	0.005 (0.01)	–0.180 (0.39)	1.302* (1.99)
Number of parental partners	–0.081 (0.68)	–0.087 (0.81)	–0.692** (2.77)
Quality of Family Relationships
Contact in prison	1.113** (3.14)	2.646** (7.08)	3.240** (4.22)
Pre-arrest support	0.180 (1.83)	0.425** (4.29)	0.373* (2.04)
Crime and Drug Use
Using drugs/alcohol	0.343 (0.98)	0.138 (0.44)	–0.644 (1.47)
Arrested	0.624 (1.23)	0.098 (0.18)	0.628 (0.97)
Restraining order	0.197 (0.61)	0.484 (1.50)	–0.300 (0.65)
Pseudo-R2		.240
Number of Children		259
Number of Child-Waves		923

Note: Regressions also control for child’s age, race and ethnicity, and fixed effects for each survey wave. Standard errors adjust for clustering by child. Absolute z statistics are shown in parentheses.

^*p < .05;

^**p < .01

Table 4.

Multinomial logistic regression analysis of contact with formerly incarcerated parents for biological children only: Boston Reentry Study

	Occasional Contact(1)	Weekly Contact(2)	Coresidence(3)
Constant	–1.668 (1.46)	–2.848* (2.51)	–6.316* (2.27)
Life Conditions After Prison Release
Unstable housing	–1.021 (1.73)	–1.559** (3.04)	–4.023** (3.56)
Reincarcerated	–3.745 (2.78)	–3.113** (3.11)	–17.844** (14.46)
Income	0.002 (0.10)	0.011 (0.60)	0.031 (1.47)
Demographics of Incarcerated Parent
Female	–1.703* (2.12)	0.280 (0.47)	3.815** (4.34)
Number of parental partners	–0.096 (0.50)	–0.283 (1.79)	–0.631** (2.69)
Quality of Family Relationships
Contact in prison	2.131** (3.38)	3.802** (6.78)	4.088 (1.61)
Pre-arrest support	–0.055 (0.34)	0.305* (2.55)	0.785** (2.73)
Feelings toward partner	0.038 (0.21)	0.277 (1.76)	0.686** (2.81)
Crime and Drug Use
Using drugs/alcohol	–0.409 (0.76)	–0.511 (1.06)	–2.687** (3.73)
Arrested	1.433 (1.81)	0.408 (0.54)	1.495 (1.44)
Restraining order	–0.577 (1.01)	0.259 (0.55)	–0.227 (0.35)
Pseudo-R2		.392
Number of Children		157
Number of Child-Waves		528

Notes: Regressions also control for child’s age, race and ethnicity, and fixed effects for each survey wave. Standard errors are clustered by child. Absolute z statistics are shown in parentheses.

^*p < .05;

^**p < .01

Patterns of post-release contact are also closely related to the complexity of family relationships. While in prison, respondents regularly said they were a parent figure to a friend or relative’s child. However, after incarceration, they were in closest contact with their own biological children. Children were more likely to be living with their mothers than their fathers, but nonresident mothers were unlikely to be in any contact at all. Finally, a parent having multiple partners is associated with reduced coresidence. Each additional parental partner is associated with a 50 % reduction in the odds of coresidence (1 − exp[−0.692] = 0.50). Notably, multiple-partner fertility is not associated with reduced weekly contact for nonresident parents; the effect is confined to coresidence.

The quality of family relationships is also closely associated with contact with children. For each point on the 5-point scale of pre-arrest parental support, the odds of regular contact with children after incarceration rise by approximately one-half. Contact between parent and child during incarceration is also strongly related to parental involvement after incarceration. In short, parents who have a history of contact and support before and during incarceration and who have a warm relationship with their co-parent are more likely to be living with their children or seeing them regularly.

Finally, the heterogeneity of formerly incarcerated parents is measured by their arrest records and other indicators of antisocial behavior. The estimates show that contact with children is not significantly related to current arrests or a record of prior domestic violence. However, parents who are using drugs or alcohol are relatively unlikely to be coresiding with children, although the estimate falls below the conventional levels for statistical significance.

When the analysis is restricted to biological children (approximately two-thirds of all children in the sample), associations between covariates and child contact become somewhat stronger. Among the measures of socioeconomic insecurity, only unstable housing is strongly related to all levels of child contact. The odds of regular contact with children for parents with unstable housing is only about one-fifth as high as the odds of child contact for stably housed parents (exp[−1.559] = 0.21). Reincarcerated parents have virtually no likelihood of being in regular contact with their biological children (exp[−4.023] = 0.02). Similar to the results for all children, parent’s income is unrelated to contact with biological children. Having a place to stay seems more important than financial means for regular contact with children, at least in the year after prison release when incomes are very low.

Estimates of the effects of family complexity for biological children are stronger than those in the full sample of biological and social children. Consistent with other research on the custodial role of formerly incarcerated parents, mothers in the reentry study are more likely than fathers to be living with their children (cf. Kruttschnitt 2010; Travis 2005:124). The estimates also indicate that mothers are highly likely to be living with their children but unlikely to be in any other sort of contact. The “visiting relationship” of formerly incarcerated parents with their partners and children describes fathers but not mothers (for fathers, see Geller 2013). As for the full sample, multiple partners are also associated with reduced odds of coresidence with biological children. This result seems to be due more to the difficulty of maintaining multiple residences than to a reduced commitment to children. Thus, respondents with more than one partner are just as likely as respondents with a single parental partner to remain in weekly contact with their biological children.

A history of family support is also related to regular contact with biological children after prison release. Parents who stay in contact with their children through visits and phone calls during imprisonment are likely to be in regular contact in the following year. Parents with positive feelings toward their parental partners are also likely to see their children regularly. Positive feelings likely depend partly on regular contact with children; nevertheless, close and regular contact with children is most likely in the presence of a congenial relationship with parental partners.

Finally, among the measures of crime and other antisocial behavior, only alcohol and drug use show the expected negative and significant associations with parental contact. The drug and alcohol coefficient for coresidence is particularly large. The estimate indicates that the odds of coresidence are 16 times higher for parents who are not using drugs or alcohol compared with those who are (exp[2.687] = 14.7). Neither arrest nor prior restraining orders are associated with reduced contact with biological children.

Qualitative Cases

The quantitative analysis describes the frequency of parental contact and associations with key predictors. However, the nature—and not just the frequency—of contact varies greatly across children. Regression coefficients, too, offer only a rough signal of the social process that unites parents with children after incarceration. The following qualitative cases illuminate the nature of parent-child contact and the social process that ties parental contact to housing, family complexity, crime, and drug use.

Qualitative data for the case studies were obtained at many different stages of data collection. We often spoke to respondents on the phone between interviews, talked to other family members, and shared informal conversations at the beginning and end of interviews. The qualitative data were richest where we completed a full set of four audiotaped follow-up interviews and an additional interview with a family member. The case studies presented in this section are for three parents—Bobby, Brian, and Carla (all pseudonyms)—who became well known to us. They provided a complete set of interviews, we spoke to others in their families, and we developed a strong rapport over the field period. These respondents span the age distribution of the sample, represent the sample’s racial diversity, and vary in their experiences of housing instability, family complexity, drug use, and criminal involvement.

Figure 2 shows a three-way cross-tabulation with the associations among measures of family complexity, housing stability, the respondent’s history of addiction. Bobby and Brian (lower panel of Fig. 2) are typical in their housing and addiction status among those with just one parental partner. Among parents with more than one parental partner, Carla (top panel of Fig. 2) is unusual for the stability of her housing. Her case illustrates the importance of housing for parental contact.

Fig. 2.

Associations between family complexity, housing stability, and addiction status among BRS parents (N = 81). Case study respondents are shown in the cells of the table. Complex families are those in which a respondent has more than one parental partner. Unstable housing is defined as ever living in unstable housing in the follow-up period. Addiction is defined as reporting a history of drug or alcohol problems at baseline

A young Puerto Rican man, Bobby was 18 and living half the time with his girlfriend and the other half with his mother, Isabel. Bobby was making a good living selling drugs, and he told us that he was driving a Lexus on his earnings. Bobby’s girlfriend was pregnant with their daughter when he was arrested. By the time he was sentenced to prison on drug charges, his daughter was 4 years old, and she had become as much a presence in Isabel’s life as in Bobby’s. Isabel had a steady public sector job and a stable home that provided the foundation for Bobby’s relationship with his daughter.

While he was in prison, Bobby kept in contact with his daughter. She had regularly stayed over with Isabel before he went to prison, and this practice continued through Bobby’s two-year incarceration. He would call his mother two or three times per week while his daughter was visiting. Father and daughter would speak on the phone several times per week. Isabel also visited Bobby in prison, and his daughter visited him in prison a few times with her grandmother.

A few months out of prison, Bobby struggled to find work but had settled into a family routine. He lived with Isabel, had a warm relationship with his daughter’s mother, and was in daily contact with his daughter, now age 6, who stayed with Isabel two nights per week. By the summer, approximately six months after his release, Bobby was taking his daughter to camp while her mother worked. With Isabel’s help, he provided a little financial support. We asked Isabel what was the best part about Bobby being out of prison. “I think it’s the bond he has with his daughter,” she said. “For me as a single parent, I feel good that I’ve taught Bobby that …. His dad was really never in his life. So to see the bond that he has with his daughter, the responsibility he feels as a father …. It makes me feel good. I showed him that. That came from me.”

With stable housing and in frequent contact with her granddaughter, Isabel helped Bobby stay in contact with his daughter’s mother and maintain a close relationship with his daughter. Unemployment left him with little income, but his mother’s house provided a venue for his daughter’s overnight visits and an informal shared custody.

Bobby’s experience was common. Many young fathers in the reentry study received housing support from older female relatives (Western et al. 2015). Stable housing facilitated contact with children by providing a safe and accessible space where families could spend time together. Approximately one-half of all parents in the sample reported stable housing over the entire year after prison. Children like Bobby’s daughter, whose parents were continuously in stable housing after prison, were 50 % more likely to be living with their parents compared with children whose parents were unstably housed.

Housing security and family support often contended with the destabilizing effects of crime and drug addiction. Many respondents we interviewed spoke of drug addictions that threatened positive relationships with their children even after periods of sobriety. Carla, a black woman in her early 40s, had used drugs since she was a teenager. Periods of heroin use, street life, and incarceration had frequently removed her from her family for months and sometimes years at a time.

The instability of Carla’s life was connected to the complexity of her family relationships. She had her first son, Reshawn, when she was 19. She and Reshawn’s father were “just messing” at the time of his birth and were not involved in a serious relationship. Carla gave birth two years later to her daughter, Jada. Jada’s father was Carla’s “soul mate,” and she continued to be involved with him for the next 20 years, although he himself was serving a long prison sentence at the time of our interviews. Three years after Jada was born, Carla had her third child, Tyrone, with a boyfriend she was living with at the time. We heard conflicting accounts of Carla’s involvement with her children. She told us that she always took Jada with her everywhere, but her sister said that all three children lived primarily with Carla’s mother, Candice, who had legally adopted the three children. Jada agreed. When we asked her, at age 21, how her mother’s drug addiction had affected her, Jada said she “didn’t really deal with it” because she lived with her grandmother, Candice.

Carla moved back in with her mother after her release from prison. She said her mother’s house was “hard” because “my kids are doing their own thing.” Reshawn, aged 23, had moved out with his girlfriend and their infant daughter. Jada, herself just out of prison, spent time at Candice’s place and her girlfriend’s. Tyrone, aged 17, lived with his grandmother but expected to move in with his girlfriend when their child was born in the following months. Carla identified strongly as her children’s mother, but she acknowledged the parental role that Candice played in her children’s lives. Upon returning to her mother’s household, Carla remarked that Candice might be “jealous of me and my kids’ relationship, because she raised them, basically.”

Echoing the regression results, Carla was little involved in her children’s lives while she was using drugs. Instead, Carla’s mother had custody. After prison, Carla lived with her mother and stopped using heroin. Similar to Bobby’s case, her mother’s place provided a setting in which she could live with her children even though she did not have custody. With different partners for each of her three children, Carla also exemplified the family complexity reported by two-thirds of the parents with at least two children. Consistent with the quantitative results, none of Carla’s children had lived with their fathers, coresiding instead with their grandmother for most of their lives.

Unlike Carla and Bobby, Brian was in a transitional housing program after prison release. A white man in his early 50s, Brian had become addicted to OxyContin after a workplace accident. Family life became chaotic as he descended into addiction. He spoke about falling into rages and blacking out. “One time I walked in the house with a gun out, and [my wife] felt like I was going to shoot her,” he said. “I should have been dead, the amount I was doing,” he said. “I don’t know how I survived it.” He was sent to prison for a string of robberies, stealing opioids from pharmacies at gunpoint.

Brian was married with two daughters who had grown into teenagers in the nine years he had been incarcerated. During his imprisonment, his wife sued for divorce and moved away from Boston. Brian wrote to his children and called them from prison. Similar to one-third of the reentry study sample, Brian moved into a transitional housing program after he was released. His program provided counseling and helped him find work. As conditions of parole, he was regularly drug-tested and attended a drug treatment program. A few months after his release, he found a job in a diner making sandwiches.

After prison, Brian made plans to reconnect with his wife and children. He had learned something of his family’s traumatic experience with his drug use. His children had nightmares and had been through counseling, he said. Living in a transitional housing program, he could not stay with them. On parole, he could not travel interstate to visit. The family finally visited Brian for the first time one year after his release. Six months later, when his conditions of supervision were relaxed, he left Massachusetts to stay with them for two weeks, the maximum time allowed. Brian kept his suitcase locked when he stayed with his family. “When you don’t have that many possessions when you are in prison, you just get used to locking things up,” he explained in our interview. “Why do you lock your suitcase?” his daughter asked. “Are you hiding drugs in there?” Brian recounted the story:

I said, “No, come here, this is the combination, and let’s go through the whole thing.” And we went through the whole suitcase, and we looked through everything, and I gave her the combination and I said, “Anytime you want to look in if I do happen to lock it, feel free to look through.” And I said, “You’re helping me. I’m helping you, and you’re helping me, but if you think I’m doing anything wrong at all, you gotta tell me right away.” So it was a pretty good moment for the both of us.

He told us that he might move closer to his family but would first have to transfer his community supervision from Massachusetts. We often heard how untreated drug addiction disabled the men and women of the reentry study from playing a positive role in their children’s lives. Parents’ drug use also created anxiety and distrust among children. Brian, sober through his first year after release, worked in a slow and deliberate way to rebuild trust with his children.

Bobby and Carla lived with their mothers in the year after incarceration, which allowed them to live with their children. Brian lived in a supportive housing program, and he could visit his children only when parole let him travel interstate.⁵ At the zenith of their addictions, Brian and Carla neglected and alienated their children. Carla’s children each had a different father, and all were adopted by Carla’s mother. Unstable housing, family complexity, crime, and drug use all tended to pull parents away from their children. Staying sober and living steadily in a private household at least provided the conditions for parents and children to come together after incarceration. Parents may share a household with their children or be in weekly contact, but the qualitative data show that family relationships can be suffused with distrust, anger, and unfamiliarity that are worn away slowly after incarceration.

Discussion

Researchers often report that parental incarceration negatively affects child well-being, yet few studies have directly observed parent-child relationships and parental involvement immediately following incarceration. Despite evidence for the negative effects of parental incarceration on children, the economic insecurity of formerly incarcerated parents and the complexity of family life were largely bracketed from prior research. We explore the contact between parents and children after incarceration using data from the Boston Reentry Study. Following a cohort of 95 formerly incarcerated parents and their 270 children reveals great variety in the involvement of parents with their children.

Three findings stand out. First, patterns of housing were closely associated with parent-child relationships after incarceration. Coresidence was uncommon, but more than one-half the sample were in weekly contact with their children. Rates of regular contact with children were significantly lower, however, for parents who were unstably housed in shelters or transitional housing or who were homeless. One-half of formerly incarcerated parents were unstably housed at some point in the year after prison release. Little analyzed in previous research but common in the lives of those who go to prison, reincarceration—even after arrest and substance use were controlled for—effectively eliminated frequent contact between parents and children.

These findings suggest that housing imposes a hard material constraint on the levels of contact between children and parents after incarceration. Although the housing effects point to economic insecurity after incarceration as an important influence on parental contact, another indicator of economic status—income—was not closely related to either coresidence or weekly contact with children. In the immediate aftermath of incarceration, unemployment rates are high, and incomes are very low. Parents’ capacity for economic contributions—important for parental contact in other research—may be so diminished just after release from prison that other dimensions of material well-being become more influential. In situations where children are not coresident but may sometimes stay with formerly incarcerated parents, stable private housing appears to be a special type of resource for promoting parent-child connections. Stable housing in private households provides not just a basic condition for social integration after prison but also a place for children to safely visit and stay overnight.

Second, the complexity of family relationships is also closely associated with parent-child contact. Family structure and relationship quality matter in different ways. The rate of multiple-partner fertility we observed in the BRS exceeded estimates from earlier studies. Formerly incarcerated parents with several partners were unlikely to live with their children, but they did maintain regular contact. Relationship quality, on the other hand, is associated with both coresidence and regular contact. Regular contact with children was much more likely for formerly incarcerated fathers and mothers who had a history of supporting their children, remained in contact during incarceration, and retained a good relationship with their parental partners. These patterns underline the gatekeeping role of partners of poor nonresident fathers observed in other research on low-income families (Edin and Nelson 2013; Tach et al. 2010, 2014). In addition to the role of partners, qualitative data suggest that relationships between formerly incarcerated parents and their children are sustained by other family members—especially older women—who played a bridging role (see also Turanovic et al. 2012). Grandmothers, in particular, provided places for visits with children and helped maintain contact between parents and children during parents’ incarceration.

Incarceration creates a distinct set of conditions for family relationships. Criminal stigma must be overcome in the labor market and social life, and institutional housing in shelters and programs often follows prison release. Still, the current findings on family complexity are anticipated by other research on disadvantaged children, not just those with incarcerated parents. The gatekeeping role of parental partners and the family support provided by older women is characteristic of poor families in general rather than those specifically with criminal justice contact (Edin and Kefalas 2005; Stack 1974). When it comes to the effects of family complexity, parent-child contact after incarceration appears to be related more to the dynamics of poverty than criminal justice.

Third, drug use and crime were associated with reduced contact with children. Researchers have previously observed the corrosive effects of drug use on family relationships among incarcerated parents (Sered and Norton-Hawk 2014; Siegel 2011). In the BRS, the scope of risky behavior and its aftermath ranged from addiction relapse for parents with long histories of substance abuse to criminal justice contact through rearrest and reincarceration. Time-varying measures of drug use and justice system contact were more closely associated with parental contact than the time-invariant prior restraining order. Addiction and criminal involvement are often viewed as fixed propensities that drive selection into incarceration (Wildeman and Turney 2014) or moderate its effects (Siegel 2011:76–93). Our results indicate that crime and drug use may be less fixed characteristics but fluctuate with the ups and downs of life after incarceration. The qualitative interviews suggest that with desistance from crime and drugs, parents can begin to rebuild positive relationships with their children.

Qualitative evidence showed that housing insecurity, family complexity, and drug use did not operate not in a linear and additive way but were often closely related. Sobriety and criminal desistance allowed formerly incarcerated parents to stay with their families. Crime and drug use often shaped the context for family complexity and conflict that tended to separate parents from children. Although regression analysis can capture the outlines, the social process of parenthood for those who go to prison is embedded in conditions of material hardship knitted with intricate kin relations.

The BRS has previously been used to study family support, employment, and rearrest (Western et al. 2015). In the current analysis, we find that family support, economic security, drug use, crime, and criminal justice contact shape how formerly incarcerated parents reunite with their children. In this way, the current analysis contributes not just to research on the wellbeing of children under mass incarceration but also to research on social integration after prison.

The social integration of formerly incarcerated parents through contact with their children depends on a rudimentary level of well-being that includes supportive family relationships and stable housing under conditions of sobriety. Good relationships with partners are formed within a wider kin network in which older female relatives play a fundamental role. Grandmothers who provide stable housing frequently help manage the relationships between their formerly incarcerated children and their partners. These findings indicate the substantial material and demographic challenges facing formerly incarcerated parents, even when they are strongly motivated to be involved in their children’s lives. Living with housing insecurity, often in the grip of addiction, and managing time and relationships with several partners are objective barriers to positive parenting for even the most committed mothers and fathers newly released from prison.

Appendix

We study the robustness of the results to alternative models drawing on an approach described by Leamer (1983), which involves recording variability in a set of coefficients of interest over all possible subsets of possible covariates. Additional covariates in the current analysis include measures of correctional supervision (time served in prison and probation or parole status), baseline measures of physical and mental health (self-reported depression, anxiety, chronic pain, and chronic disease), a measure of family history (growing up in a single-parent family), and a measure of the transition from prison to community (social isolation in the first week after prison release). Means of covariates are reported in Table 5. The eight candidate covariates are correlated with the covariates of interest and are plausibly related to parental contact after incarceration. With eight candidate covariates, analysis explores sensitivity of the coefficients over 2⁸ = 256 possible models.

Sensitivity of the results is measured in two ways. First, we report the proportion of regression coefficients of 256 alternative models with the opposite sign of that reported in Tables 3 or 4. Leamer (1983) initially proposed a change in sign as an indication of sensitivity to the model specification. Second, we report the proportion of nonsignificant coefficients over the 256 alternative models.

Results of the sensitivity analysis are reported in Table 6. Lower numbers close to 0 indicate greater stability of the reported regression results across alternative models. Reading across the first row of the table, for example, the coefficient for unstable housing never changes sign across 256 alternative models. Unstable housing is associated with less contact with children across 256 alternative models. The second row indicates that the unstable housing coefficient is never significant for the “occasional contact” outcome but is always significant for weekly contact and coresidence. Results are generally robust to a wide range of alternative models for the key coefficients for unstable housing, reincarceration, parental contact with a biological child, contact in prison, pre-arrest support, feelings toward partner, and drug and alcohol use.

Table 5.

Means of covariates used in sensitivity test

	Sample of Children
	Biological	Nonbiological	All
Length of Prison Stay (months)	30.285	30.755	30.452
Post-release Supervision	0.624	0.642	0.631
Depression	0.391	0.477	0.421
Anxiety	0.129	0.138	0.132
Chronic Pain	0.352	0.361	0.355
Infectious/Chronic Disease	0.339	0.254	0.309
Living With Both Parents at Age 14	0.284	0.159	0.239
Social Isolation After Release (z score)	−0.119	−0.578	−0.279

Note: All covariates are dummy variables unless otherwise indicated.

Table 6.

Proportion of multinomial logit coefficients changing sign and proportion of 95 % intervals including 0, over 256 alternative models

	All Children			Biological Children
	(1)	(2)	(3)	(4)	(5)	(6)
Unstable Housing	.00	.00	.00	.00	.00	.00
	1.00	.00	.00	1.00	.00	.00
Reincarcerated	.00	.00	.00	.00	.00	.00
	.50	.00	.00	.01	.00	.00
Income	.03	.00	.00	.02	.00	.00
	1.00	1.00	1.00	1.00	1.00	1.00
Biological Child	.00	.00	.00	––	––	––
	1.00	.00	.99	––	––	––
Female	.43	.25	.00	.00	.00	.00
	1.00	1.00	.16	.00	.58	.00
Number of Parental Partners	.03	.00	.00	.50	.00	.00
	1.00	.87	.00	1.00	.49	.14
Contact in Prison	.00	.00	.00	.00	.00	.00
	.00	.00	.00	.00	.00	.16
Pre-arrest Support	.00	.00	.00	.00	.00	.00
	.73	.00	.00	1.00	.00	.00
Feelings Toward Partner	––	––	––	.25	.00	.00
	––	––	––	.00	.00	.00
Using Drugs/Alcohol	.00	.13	.00	.01	.00	.00
	1.00	1.00	.97	1.00	1.00	.00
Arrested	.00	.31	.00	.00	.00	.00
	1.00	1.00	1.00	.98	1.00	1.00
Restraining Order	.00	.00	.16	.00	.00	.00
	1.00	1.00	1.00	1.00	1.00	1.00

Notes: Columns 1–3 correspond to coefficients reported in Table 3. Columns 4–6 correspond to coefficients reported in Table 4. Sensitivity of the coefficients was examined for baseline models reported in Tables 3 and 4, plus all possible subsets of the following predictors: time served, probation or parole, depression, anxiety, chronic pain, chronic disease, growing up in a single-parent household, and social isolation in the first week after prison release.

Table 7.

Multinomial logistic regression analysis of contact with formerly incarcerated fathers for all biological and social children: Boston Reentry Study

	Occasional Contact (1)	Weekly Contact (2)	Coresidence (3)
Constant	–2.070* (2.02)	–3.665** (3.53)	–3.322 (1.92)
Life Conditions After Prison Release
Unstable housing	–0.123 (0.37)	–0.712* (2.53)	–1.130* (2.26)
Reincarcerated	–1.817* (2.21)	–2.221** (3.10)	–14.979** (18.01)
Income	–0.005 (0.33)	0.012 (1.04)	0.017 (1.14)
Demographics of Incarcerated Parent
Biological child	0.824* (2.05)	1.515** (4.08)	–0.382 (0.74)
Number of partners	–0.105 (0.56)	–0.006 (0.03)	–0.132 (0.33)
Quality of Family Relationships
Contact in prison	0.936* (2.54)	2.671** (6.29)	4.389** (3.42)
Pre-arrest support	0.147 (1.28)	0.437** (3.56)	0.430* (2.22)
Crime and Drug Use
Using drugs/alcohol	0.563 (1.44)	0.213 (0.58)	–0.689 (1.35)
Arrested	0.868 (1.71)	0.326 (0.62)	–0.107 (–0.13)
Restraining order	0.044 (0.12)	0.296 (0.89)	–1.131* (2.30)
Pseudo-R2		.258
Number of Children		214
Number of Child-Waves		753

Notes: Regressions also control for child’s age, race and ethnicity, and fixed effects for each survey wave. Standard errors are clustered by child. Absolute z statistics are shown in parentheses.

^*p < .05;

^**p < .01