hopelessly devoted? relationship quality during and after ... · 484 journalofmarriageandfamily...

16
Kristin Turney University of California, Irvine Hopelessly Devoted? Relationship Quality During and After Incarceration A growing literature highlights the multifaceted consequences of incarceration for family life, but little is known about the quality of relationships between couples who remain together during and after 1 partner’s incarceration. In this article, the author used data from the Fragile Families and Child Wellbeing Study (N = 1,848), a longitudinal cohort of parents, to consider the association between paternal incarceration and 4 measures of relationship quality: (a) overall relationship quality, (b) supportiveness, (c) emotional abuse, and (d) physical abuse. The results showed that paternal incarceration in the past 2 years was, by and large, asso- ciated with lower mother-reported (but not father-reported) relationship quality. However, across some outcome variables, current pater- nal incarceration was positively associated with relationship quality. Taken together, these findings suggest that current and recent incar- ceration have countervailing consequences for relationship quality and, more generally, that the penal system exerts a powerful influence even among couples who maintain relationships. The dramatic rise in incarceration in the United States, as well as its consequences for offenders and those connected to them, is by now well known (Sampson, 2011; Wakefield & Uggen, Department of Sociology, University of California, Irvine, 3151 Social Science Plaza, Irvine, CA 92697–5100 ([email protected]). This article was edited by Robert Crosnoe. Key Words: Fragile Families and Child Wellbeing Study, incarceration, relationship quality. 2010). The 2.3 million individuals currently incarcerated, as well as the many others released annually from prisons and jails, are not solitary individuals but are instead connected to fam- ily members as romantic partners and parents (Glaze, 2011). Given the family roles inhab- ited by currently and formerly incarcerated individuals, the majority of whom are men, it is unsurprising that mass incarceration has wide-ranging collateral consequences for fam- ily life (for reviews, see Wakefield & Uggen, 2010, and Wildeman & Muller, 2012). Perhaps most substantially, a relatively large literature documents that incarceration increases the risk of marital dissolution (Apel, Blokland, Nieuwbeerta, & van Schellen, 2010; Lopoo & Western, 2005; Massoglia, Remster, & King, 2011; Western, 2006; also see Geller, 2013). But the focus on marital dissolution leaves a number of answered questions about the col- lateral consequences of incarceration for family life. For one, although many individuals behind bars are in nonmarital romantic relationships, relatively few of the incarcerated are in marital unions (Western, Lopoo, & McLanahan, 2004). Therefore, any examination of incarceration’s consequences for marital dissolution is inap- plicable to a large segment of the incarcerated population, potentially underestimating the familial consequences of incarceration, and it may be at least equally informative to examine incarceration’s consequences for relationship quality among (marital and nonmarital) roman- tic partners. On a related note, incarceration does not unequivocally lead to relationship dissolution (e.g., Comfort, 2008), and under- standing the quality of the relationships among couples who remain together may provide broad 480 Journal of Marriage and Family 77 (April 2015): 480–495 DOI:10.1111/jomf.12174

Upload: others

Post on 22-Oct-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

  • Kristin Turney University of California, Irvine

    Hopelessly Devoted? Relationship Quality During

    and After Incarceration

    A growing literature highlights the multifacetedconsequences of incarceration for family life, butlittle is known about the quality of relationshipsbetween couples who remain together duringand after 1 partner’s incarceration. In thisarticle, the author used data from the FragileFamilies and Child Wellbeing Study (N= 1,848),a longitudinal cohort of parents, to considerthe association between paternal incarcerationand 4 measures of relationship quality: (a)overall relationship quality, (b) supportiveness,(c) emotional abuse, and (d) physical abuse.The results showed that paternal incarcerationin the past 2 years was, by and large, asso-ciated with lower mother-reported (but notfather-reported) relationship quality. However,across some outcome variables, current pater-nal incarceration was positively associatedwith relationship quality. Taken together, thesefindings suggest that current and recent incar-ceration have countervailing consequences forrelationship quality and, more generally, that thepenal system exerts a powerful influence evenamong couples who maintain relationships.

    The dramatic rise in incarceration in the UnitedStates, as well as its consequences for offendersand those connected to them, is by now wellknown (Sampson, 2011; Wakefield & Uggen,

    Department of Sociology, University of California, Irvine,3151 Social Science Plaza, Irvine, CA 92697–5100([email protected]).

    This article was edited by Robert Crosnoe.

    Key Words: Fragile Families and Child Wellbeing Study,incarceration, relationship quality.

    2010). The 2.3 million individuals currentlyincarcerated, as well as the many others releasedannually from prisons and jails, are not solitaryindividuals but are instead connected to fam-ily members as romantic partners and parents(Glaze, 2011). Given the family roles inhab-ited by currently and formerly incarceratedindividuals, the majority of whom are men,it is unsurprising that mass incarceration haswide-ranging collateral consequences for fam-ily life (for reviews, see Wakefield & Uggen,2010, and Wildeman & Muller, 2012). Perhapsmost substantially, a relatively large literaturedocuments that incarceration increases therisk of marital dissolution (Apel, Blokland,Nieuwbeerta, & van Schellen, 2010; Lopoo &Western, 2005; Massoglia, Remster, & King,2011; Western, 2006; also see Geller, 2013).

    But the focus on marital dissolution leavesa number of answered questions about the col-lateral consequences of incarceration for familylife. For one, although many individuals behindbars are in nonmarital romantic relationships,relatively few of the incarcerated are in maritalunions (Western, Lopoo, & McLanahan, 2004).Therefore, any examination of incarceration’sconsequences for marital dissolution is inap-plicable to a large segment of the incarceratedpopulation, potentially underestimating thefamilial consequences of incarceration, and itmay be at least equally informative to examineincarceration’s consequences for relationshipquality among (marital and nonmarital) roman-tic partners. On a related note, incarcerationdoes not unequivocally lead to relationshipdissolution (e.g., Comfort, 2008), and under-standing the quality of the relationships amongcouples who remain together may provide broad

    480 Journal of Marriage and Family 77 (April 2015): 480–495DOI:10.1111/jomf.12174

  • Incarceration and Relationship Quality 481

    insight into family functioning that has beenoverlooked in studies of marital dissolution.

    It is not immediately obvious whether incar-ceration will be deleterious, beneficial, orinconsequential for relationship quality amongcouples who maintain their romantic relation-ships. On the one hand, there are a numberof challenges associated with maintaining arelationship during or after incarceration—including lack of shared time together, the eco-nomic costs of maintaining contact, and the emo-tional toll experienced by both partners—thatmay strain romantic relationships (e.g.,Comfort, 2008; Massoglia et al., 2011). Onthe other hand, qualitative research suggests thatthe time spent apart during one partner’s incar-ceration allows couples to find their relationshipstride in ways not possible outside of the prisonwalls (e.g., Braman, 2004; Comfort, 2008).Alternatively, because the incarcerated are nota random slice of the population and are insteaddisadvantaged across an array of characteristics,it is also possible that incarceration has no inde-pendent association with relationship quality.

    In this study, I considered these possibilitieswith data from the Fragile Families and ChildWellbeing Study (FFCWB; see http://www.fragilefamilies.princeton.edu/), a longitudinalcohort of new parents and their children, toprovide the first systematic quantitative exami-nation of the association between incarcerationand relationship quality among couples whoremain together despite incarceration. Under-standing the potentially complex associationbetween incarceration and relationship qualityis important given that high-quality roman-tic relationships are positively associated withhealth (e.g., House, Landis, & Umberson, 1988),parenting (e.g., Grych & Fincham, 1990), andrelationship longevity (e.g., Gottman, 1994)among adults and positively associated withchild well-being (e.g., Crosnoe & Cavanagh,2010; Grych & Fincham, 1990). Furthermore,research suggests that parental relationships maybe one mechanism through which paternal incar-ceration exerts deleterious effects on children(Geller, Cooper, Garfinkel, Schwartz-Soicher,& Mincy, 2012).

    Background

    Mass Incarceration and Family Stress Theory

    Family stress theory suggests that stressfulevents generate transformations to the family

    system (McCubbin, 1979). Family stress, espe-cially nonnormative stress such as incarceration,may cause families to disintegrate and dete-riorate (Lavee, McCubbin, & Olson, 1987).Indeed, incarceration is a stressor for families(Patterson, 2002; Pearlin, 1989). Prior to incar-ceration, men contribute both economic andsocial resources to family life (Wildeman,Schnittker, & Turney, 2012). Romantic part-ners who endure incarceration along with theincarcerated—albeit outside of the prison walls—may experience stigma (Braman, 2004), eco-nomic hardship (Schwartz-Soicher, Geller, &Garfinkel, 2011), poor mental health (Wildemanet al., 2012), and reduced social support (Turney,Schnittker, & Wildeman, 2012), all of whichmay create friction and vulnerability in relation-ship stability and quality. Also, upon release, thestress of reintegration into family life may yieldfurther deterioration. Given that these marginalmen are connected to families—before, dur-ing, and after release—a burgeoning literaturehas documented the collateral and unintendedconsequences of incarceration for family life(Wildeman & Muller, 2012).

    Consistent with family stress theory, relation-ship dissolution is one of the most commonlyconsidered familial consequences of incarcer-ation. There are many reasons to expect whymaintaining a relationship with a currently orformerly incarcerated romantic partner is diffi-cult. Incarcerated men are forcefully removedfrom their households. Contact with incarceratedmen is expensive and time consuming, whichis to say nothing of the dignity women losewhile visiting prisons (Comfort, 2008), and thelack of shared time together may strain rela-tionships (Rindfuss & Stephen, 1990). Indeed,an array of quantitative research studies, mostlyusing data from the National Longitudinal Studyof Youth 1979, have found that incarcerationwas associated with an increased risk of divorce(e.g., Lopoo & Western, 2005; Massoglia et al.,2011; Western, 2006; for research on disso-lution among married and cohabiting couples,see Western et al., 2004). But not all couplesare in marital unions, and not all couples dis-solve their unions during or after incarceration.Accordingly, considering the relationship qual-ity of the marital and nonmarital couples whoremain together might provide a broad under-standing of how incarceration affects familyfunctioning.

  • 482 Journal of Marriage and Family

    Deleterious Consequences of Incarcerationfor Relationship Quality

    In accordance with family stress theory, thereare good reasons to expect that incarcerationis deleterious for relationship quality. To begin,incarceration’s influence on economic hardshipmay play a role. It is well known that incar-ceration has damaging economic consequences(e.g., Western, 2006). Prior to incarceration,most men are employed, and nearly all incarcer-ated men—and their households—experienceincome loss during incarceration. Upon release,the resultant stigma, discrimination, and loss ofhuman and social capital makes finding employ-ment difficult, which may place financial strainon family life. Indeed, family stress theory high-lights how financial strain, through its facili-tation of strained marital interactions, reducesrelationship quality (Conger et al., 1990).

    In addition, the association between incar-ceration and relationship quality may operatethrough reduced physical and mental health ofboth partners. For men who experience con-finement, regimentation, and identity transfor-mations associated with their time behind bars(Arditti, Smock, & Parkman, 2005), incarcera-tion has negative and lasting effects on mentalhealth (Schnittker, Massoglia, & Uggen, 2012).Incarceration may socialize men to engage inviolent behaviors (Nurse, 2002). Furthermore,the women in relationships with incarcerated andformerly incarcerated men also experience resul-tant distress associated with their confinement(Lowenstein, 1984; Wildeman et al., 2012). Themental health challenges do not dissipate upona man’s release, because reintegration may beespecially anxiety producing and challenging,especially if the woman suspects he will go backto prison (Goffman, 2009). Men’s participationin illegal activities or return to substance abuse,may crush the high expectations that womenhave during the incarceration period (Braman,2004). In turn, impaired mental health associ-ated with imprisonment may reduce relationshipquality (e.g., Booth & Johnson, 1994).

    Additional Possibilities

    Although incarceration may be a stressor tothe family system and have deleterious conse-quences for relationship quality among coupleswho remain together, it is also possible thatincarceration is beneficial or inconsequential for

    relationship quality. Qualitative research, espe-cially, provides a nuanced understanding of howthese couples thrive. For example, some menuse their time behind bars as an occasion toreflect on their roles as romantic partners andfathers (Edin, Nelson, & Paranal, 2004). Thisreflection often leads to a recommitment to fam-ily life and, if men express these feelings totheir romantic partners, their partners may alsobe hopeful for the future (Braman, 2004; Roy& Dyson, 2005). In Braman’s (2004) ethno-graphic account of incarceration and family life,he described the behavior of a respondent’s part-ner while in prison: “He was promising to reformhis ways, writing long letters of regret, talkingabout his religious reform in prison, and suggest-ing that they get married on his release” (p. 47).In addition, because men often stop or dramati-cally reduce their drug and alcohol use while inprison, partners—whose relationships were onceburdened with substance abuse—remain opti-mistic that these men have turned over a newleaf and, therefore, remain by their side (Bra-man, 2004; also see Comfort, 2008). Further-more, the secondary incarceration experiencedby some women—as they learn to navigate andadapt to the correctional environment, especiallywhen visiting their partners—may help women,many of whom share children with these men,preserve family relationships (Comfort, 2008).Women, especially, may stay in relationshipsbecause of pressure from family members togive men another chance or because they areacutely aware of the shortage of marriageablemen in their communities (Braman, 2004).

    Finally, it is also possible that incarcerationhas no independent effect on relationship qual-ity. In the United States, incarceration is notrandomly distributed across the population butinstead is especially concentrated among dis-advantaged groups. For example, incarceratedindividuals, compared to their counterparts, areless likely to be in marital relationships. Theyalso have lower socioeconomic status, morehealth problems, and less self-control (e.g.,Wildeman et al., 2012, pp. 237–238). The samefactors associated with incarceration may alsobe associated with poor relationship quality,and therefore it is quite plausible that any effectof incarceration on relationship quality—or onfamily life, more broadly—results not fromincarceration but from characteristics associatedwith selection into incarceration.

  • Incarceration and Relationship Quality 483

    Additional Characteristics Associated WithIncarceration and Relationship Quality

    One way to account for nonrandom selectioninto incarceration is to adjust for a host of char-acteristics that may render spurious the asso-ciation between incarceration and relationshipquality. Therefore, the analyses adjusted for ahost of demographic, relationship, economic,and health characteristics that were associatedwith both incarceration and relationship qual-ity. Demographic characteristics included race,immigrant status, age, and childhood familystructure, given that previous research showsall were associated with incarceration and rela-tionship quality (Bradbury, Fincham, & Beach,2000; Brown & Booth, 1996; Glenn, 1990;Wakefield & Uggen, 2010; Webster, Orbuch,& House, 1995). The analyses also controlledfor an array of pre-incarceration relationshipcharacteristics, such as relationship type (Brown& Booth, 1996), duration (Brown & Booth,1996), joint children (Kurdek, 1989), and atti-tudes about marriage and gender (Amato &Rogers, 1999; Karney & Bradbury, 1995; Waller& McLanahan, 2005). Economic characteristics(Conger et al., 1990; Hardie & Lucas, 2010;Wakefield & Uggen, 2010) and health charac-teristics (Booth & Johnson, 1994; Wakefield &Uggen, 2010) were associated with both incar-ceration and relationship quality.

    Method

    Data Source

    I examined the association between paternalincarceration and relationship quality with datafrom the FFCWB, a longitudinal study of par-ents born in urban areas between February 1998and September 2000. Mothers and fathers werefirst interviewed when their children were bornand were reinterviewed when their children wereapproximately 1, 3, 5, and 9 years old. Formore information about the study design andresponse rates, see Reichman, Teitler, Garfinkel,and McLanahan (2001).

    The FFCWB data were limited to a sample ofparents and were thus not a representative sam-ple of adults in romantic relationships, but theywere ideal to consider the association betweenincarceration and relationship quality. First,because the data included an oversample ofunmarried parents, a disproportionately disad-vantaged group, they comprised a relatively

    large number of previously incarceratedmen, some of whom were in stable romanticrelationships. The data also included multiplemeasures of relationship quality reported byboth partners. In addition, although the datawere observational and were therefore limitedin their ability to provide causal estimates,their longitudinal nature allowed for variousstrategies to reduce unobserved heterogeneity,as discussed below. Finally, examining relation-ship quality among parents, specifically, wasespecially important because these householdsincluded already vulnerable children for whompoor relationship quality may be especiallyconsequential (e.g., Grych & Fincham, 1990).

    The two analytic samples, one for mothers’reports of relationship quality (N = 1,848) andone for fathers’ reports of relationship quality(N = 1,585), relied on data through the 5-yearsurvey because paternal incarceration was mostprecisely measured between the 3- and 5-yearsurveys. In constructing both analytic samples,I first deleted the 1,051 observations (21%)in which the mother did not participate in the3- or 5-year surveys and the additional 1,997(41%) observations in which the parents werenot in a relationship with each other at the5-year survey. I then deleted an additional twomothers (

  • 484 Journal of Marriage and Family

    variables—excluding mother’s parenting stress,father’s parenting stress, and father’s impulsiv-ity, which had about 20% of observations withmissing data—were missing fewer than 4% ofobservations. I preserved missing observationswith multiple imputation.

    The data in Table 1 show that the sample wasrelatively disadvantaged across a wide range ofdemographic characteristics (though, as notedabove, was less disadvantaged than the fullFFCWB sample). More than 70% of motherswere racial/ethnic minorities. More than half ofmothers (55%) did not have education beyondhigh school. At the 1-year survey, about 43%of couples were married, 39% were cohabiting,16% were in a nonresidential romantic rela-tionship, and 3% were not in a relationship.Couples, on average, had known each other fornearly 6 years prior to the birth of their child,and nearly half (47%) had additional childrentogether. More than one quarter (27%) of fathersin the analytic sample were incarcerated prior tothe 3-year survey.

    Measures

    Dependent variables. The primary outcomevariables included four indicators of mothers’

    Table 1. Descriptive Statistics of Variables Used inAnalyses (N= 1,848)

    Variables % or M SD

    Mother-reported relationship qualityOverall relationship quality (range: 1–5) 3.95 0.98Supportiveness (range: 1–3) 2.61 0.40Emotional abuse (range: 1–3) 1.12 0.21Physical abuse (range: 1–3) 1.04 0.15

    Father-reported relationship qualityOverall relationship quality (range: 1–5) 4.15 0.88Supportiveness (range: 1–3) 2.69 0.35Emotional abuse (range: 1–3) 1.16 0.24Physical abuse (range: 1–3) 1.07 0.21

    Key independent variablesFather recent incarceration 6.2%Father current incarceration 2.1%

    Control variablesRace/ethnicity

    Non-Hispanic White 29.1%Non-Hispanic Black 35.1%Hispanic 31.0%Non-Hispanic other race 4.8%

    Table 1. Continued

    Variables % or M SD

    Mother and father mixed-race couple 12.4%Mother immigrant 22.2%Mother age 26.72 6.15Mother lived with both parents at 15 53.7%Mother and father relationship status

    Married 43.0%Cohabiting 38.5%Nonresidential romantic 15.7%No relationship 2.8%

    Relationship duration (years) 5.80 4.90Mother and father have additional children 46.6%Relationship commitment 2.43 0.58Mother pro-marriage attitudes 2.66 0.42Mother traditional attitudes 2.10 0.65Mother gender distrust 1.96 0.56Mother education

    Less than high school 27.6%High school diploma or GED 27.2%More than high school 45.2%

    Mother employment 70.8%Mother income-to-poverty ratio 2.97 2.94Mother depression 12.2%Father depression 7.1%Mother parenting stress 2.12 0.66Father parenting stress 2.04 0.65Mother fair or poor health 11.5%Father fair or poor health 8.9%Mother substance abuse 6.5%Father substance abuse 29.8%Mother reports domestic violence 2.5%Father impulsivity 1.91 0.64Father prior incarceration 27.0%Mother overall relationship qualitya 3.99 0.99Mother supportivenessa 2.64 0.37Mother emotional abusea 1.14 0.22Mother physical abusea 1.04 0.17Father overall relationship qualitya 4.11 0.96Father supportivenessa 2.69 0.34Father emotional abusea 1.19 0.24Father physical abusea 1.09 0.25

    Note: Data are from the Fragile Families and Child Well-being Study. The sample was limited to parents in a romanticrelationship at the 3- and 5-year surveys. Recent paternalincarceration signifies that the father was incarcerated afterthe 3-year survey and up to and including the 5-year survey(when relationship quality was measured). All control vari-ables were measured at the baseline or 1-year surveys. Ns forfather-reported relationship quality are smaller (N = 1,585)because fewer fathers than mothers participated in the 5-yearsurvey.

    aLagged.

  • Incarceration and Relationship Quality 485

    and fathers’ relationship quality at the 5-yearsurvey: (a) overall relationship quality, (b) sup-portiveness, (c) emotional abuse, and (d) physi-cal abuse. First, overall relationship quality wasmeasured with an ordinal variable (1= poor to5= excellent).

    Second, supportiveness was measured byaveraging the following items asked of moth-ers and fathers (range: 1= never to 3= often):(a) [mother/father] is fair and willing to com-promise when you have a disagreement, (b)[mother/father] expresses affection or love foryou, (c) [mother/father] encourages or helpsyou to do things that are important to you, (d)[mother/father] listens when you need some-one to talk to, and (e) [mother/father] reallyunderstands your hurts and joys (𝛼s= .76 and.70 for mothers and fathers at the 5-year survey,respectively). A principal-components factoranalysis showed that these individual itemsloaded onto the same factor.

    Third, mothers and fathers were each asked11 questions about abuse in their relationship.Emotional abuse was measured as an averageof the following seven items (range: 1= neverto 3= often): (a) [mother/father] insults orcriticizes you or your ideas; (b) [mother/father]tries to keep you from seeing or talking withyour friends or family; (c) [mother/father] triesto prevent you from going to work or school;(d) [mother/father] withholds money, makesyou ask for money, or takes your money; (e)[mother/father] tries to make you have sexor do sexual things you don’t want to do; (f)[mother/father] withholds sex or tries to controlyour behavior; and (g) [mother/father] insults orcriticizes you for not taking good enough careof the child or your home (𝛼s= .67 and .69 formothers and fathers).

    Physical abuse was measured as an aver-age of the following four items (1= never to3= often): (a) [mother/father] slaps or kicks you;(b) [mother/father] hits you with a fist or anobject that could hurt you; (c) [mother/father]throws something at you; and (d) [mother/father]pushes, grabs, or shoves you (𝛼s= .65 and .60for mothers and fathers). The two scales wereinformed by a principal-components factor anal-ysis. One item ([mother/father] tries to make youhave sex or do sexual things you don’t wantto do]) loaded onto both the emotional abuseand physical abuse scales. The factor loadingwas slightly stronger if included in the emo-tional abuse measure, though results remain

    unchanged if I instead included it in the physicalabuse measure.

    On average, parents reported high overallrelationship quality at the 5-year survey (3.95among mothers and 4.15 among fathers, on ascale of 1 to 5). Parents also reported high sup-portiveness (2.61 and 2.69, respectively, on ascale of 1 to 3), low emotional abuse (1.04 and1.16, respectively, on a scale of 1 to 3), and lowphysical abuse (1.04 and 1.07, respectively, on ascale of 1 to 3).

    Independent variables. The two independentvariables included (a) recent paternal incar-ceration and (b) current paternal incarceration.Recent paternal incarceration was coded affir-matively if the father spent time in prison orjail at any point after the 3-year survey up toand including the 5-year survey. Consistentwith other research suggesting individuals mayunderreport incarceration (e.g., Geller et al.,2012), fathers were considered recently incar-cerated if either parent reported the father wasincarcerated during this time period. Currentpaternal incarceration, which is necessary toaccount for the fact that some measures ofrelationship quality may not apply to fatherscurrently incarcerated, indicated that the fatherwas in prison or jail at the 5-year survey. Inthe analytic sample, about 6% and 2% offathers were recently and currently incarcerated,respectively.

    Control variables. The multivariate analysesadjusted for an array of individual-level char-acteristics associated with incarceration andrelationship quality. It is important to note thatall variables were measured at the baselineor 1-year surveys and, therefore, prior to themeasure of recent paternal incarceration. Demo-graphic control variables included mothers’race (non-Hispanic White, non-Hispanic Black,Hispanic, non-Hispanic other race), as wellas dummy variables indicating the mother andfather were a mixed-race couple, the mother wasforeign born, and the mother lived with bothparents at age 15. Mothers’ age was representedby a continuous variable. The analyses alsoadjusted for an array of relationship charac-teristics, including baseline relationship status(married, cohabiting, nonresidential romantic,no relationship), relationship duration, relation-ship commitment (sum of mothers’ baselinereports of the following: [a] father provided

  • 486 Journal of Marriage and Family

    financial support during pregnancy, [b] the childwill have fathers’ last name, and [c] the fathervisited the mother in the hospital; see Tach &Edin, 2013), promarriage attitudes (e.g., “It isbetter for a couple to get married than to just livetogether,” ranging from 1 [strongly disagree] to4 [strongly agree]; 𝛼 = .60), traditional attitudes(e.g., “The important decisions in the familyshould be made by the man of the house,”ranging from 1 [strongly disagree] to 4 [stronglyagree]; 𝛼 = .60), and gender distrust (e.g., “Mencannot be trusted to be faithful,” ranging from1 [strongly disagree] to 4 [strongly agree];𝛼 = .64). A dummy variable indicated that theparents shared additional children besides thefocal child. The analyses also adjusted for eco-nomic characteristics, including education (lessthan high school, high school or GED, more thanhigh school); employment in the past week; andincome-to-poverty ratio. The analyses adjustedfor health characteristics, including depression,parenting stress, and fair or poor health. Finally,the analyses adjusted for characteristics espe-cially associated with incarceration, includingfathers’ engagement in domestic violence, sub-stance abuse, impulsivity (e.g., “I will often saywhatever comes into my head without thinkingfirst,” ranging from 1 [strongly disagree] to 4[strongly agree]; 𝛼 = .83), and prior incarcera-tion (any incarceration at or before the 3-yearsurvey).

    Analytic Plan

    The analytic strategy was straightforward. I firstdocumented bivariate differences in parents’relationship quality by paternal incarceration;these data are shown in Table 2. The multivariateanalyses, presented in Tables 3 and 4, proceededin three stages. Because the analytic sampleincluded only couples in a relationship (asrelationship quality was relevant only for thesecouples), I used a first-stage logistic regressionmodel to estimate the propensity for being in arelationship at the 5-year survey as a functionof an array covariates, all measured prior to the3-year survey. Essentially, this model estimatedeach couples’ probability of being included inthe analysis (see Appendix Table A1). Second,I saved this propensity score. Third, I usedordinary least squares (OLS) regression modelsto estimate relationship quality as a functionof recent paternal incarceration, weighting theanalyses by the inverse probability of treatment

    Table 2. Means of Relationship Quality, by RecentPaternal Incarceration (N= 1,848)

    Recent paternal incarceration

    VariableYes

    (n= 114)No

    (n= 1,734) Cohen’s d

    Mother-reportedrelationship qualityOverall relationship

    quality3.35 3.99 0.66∗∗∗

    Supportiveness 2.50 2.62 0.32∗∗

    Emotional abuse 1.20 1.12 −0.38∗∗∗

    Physical abuse 1.12 1.03 −0.56∗∗∗

    Father-reportedrelationship qualityOverall relationship

    quality3.82 4.17 0.40∗∗∗

    Supportiveness 2.60 2.70 0.28∗

    Emotional abuse 1.26 1.16 −0.45∗∗∗

    Physical abuse 1.18 1.07 −0.53∗∗∗

    Note: Data are from the Fragile Families and ChildWellbeing Study. Ns for father-reported relationshipquality are smaller (85 fathers experienced recent incar-ceration and 1,500 fathers did not experience recentincarceration) because fewer fathers than mothers partici-pated in the 5-year survey. Asterisks represent statisticallysignificant differences between the two groups. ∗p< .05.∗∗p< .01. ∗∗∗p< .001 (two-tailed tests).

    (with the “treatment” being the probabilityof being in the analysis). Ordered logisticregression models may be more appropriate forestimates of overall relationship quality, whichwas not normally distributed, but supplementalanalyses produced results that were substan-tively similar to the OLS results. Therefore, forease of interpretation, I used OLS regressionacross all outcome variables.

    Because there were substantial threats tocausal inference, as discussed above, these OLSmodels took steps to account for unobservedheterogeneity between couples who did and didnot experience paternal incarceration. I adjustedfor a wide array of covariates—all measuredprior to the measurement of incarceration—thatmay render the association between incarcera-tion and relationship quality spurious (includinga lagged dependent variable). In additionalanalyses, I restricted the sample to those couplesmost at risk of experiencing paternal incarcera-tion: those who had previously experiencedpaternal incarceration. In this model, the

  • Incarceration and Relationship Quality 487

    Table 3. Ordinary Least Squares Regression Models Estimating Mother-Reported Relationship Quality (N= 1,848)

    Overall relationshipquality Supportiveness Emotional abuse Physical abuse

    Variable 𝛽 SE 𝛽 SE 𝛽 SE 𝛽 SE

    Father recent incarceration −0.14 0.18∗∗ −0.15 0.07∗ 0.02 0.04 0.12 0.03†Father current incarceration 0.06 0.32 0.11 0.10∗ −0.05 0.05 −0.12 0.04∗∗Race/ethnicitya

    Non-Hispanic Black −0.09 0.09∗ 0.00 0.03 0.01 0.02 0.05 0.02Hispanic −0.02 0.09 0.03 0.03 −0.04 0.02 −0.05 0.01Non-Hispanic other race 0.02 0.17 0.03 0.06 −0.04 0.04 −0.04 0.02

    Mother and father mixed-race couple −0.04 0.15 0.01 0.03 0.06 0.04 0.13 0.04†Mother immigrant 0.01 0.10 0.00 0.03 0.06 0.02∗ 0.04 0.02Mother age −0.03 0.01 −0.02 0.00 0.04 0.00 0.04 0.00Mother lived with both parents at 15 −0.04 0.08 −0.03 0.02 0.02 0.02 0.06 0.02Relationship statusb

    Cohabiting −0.07 0.10† −0.03 0.04 0.09 0.02∗ 0.06 0.02Nonresidential romantic −0.05 0.11 0.00 0.04 0.01 0.02 −0.01 0.02No relationship −0.07 0.19 −0.05 0.07 0.07 0.05 0.10 0.04

    Relationship duration 0.01 0.01 −0.03 0.00 −0.08 0.00∗∗ −0.08 0.00∗Mother and father have more children 0.02 0.08 −0.05 0.03 0.03 0.02 0.03 0.01Relationship commitment 0.07 0.08 −0.03 0.03 −0.03 0.02 −0.05 0.01Mother pro-marriage attitudes −0.05 0.10 −0.02 0.04 0.09 0.02∗ 0.10 0.02∗Mother traditional attitudes 0.04 0.05 0.00 0.02 −0.02 0.01 −0.04 0.01Mother gender distrust −0.03 0.07 −0.05 0.02† 0.04 0.01 0.03 0.01Mother educationc

    High school diploma or GED −0.02 0.09 −0.04 0.04 0.03 0.02 0.00 0.02More than high school 0.00 0.10 −0.04 0.04 0.01 0.02 −0.09 0.02†

    Mother employment 0.00 0.08 −0.02 0.03 0.02 0.02 0.01 0.01Mother income-to-poverty ratio −0.01 0.02 0.01 0.00 0.06 0.00 0.08 0.00Mother depression 0.00 0.11 −0.04 0.05 0.02 0.03 0.00 0.02Father depression 0.03 0.15 0.00 0.05 0.01 0.03 0.03 0.02Mother parenting stress 0.01 0.05 −0.04 0.02 −0.02 0.01 −0.03 0.01Father parenting stress −0.01 0.05 0.06 0.02† −0.02 0.01 −0.03 0.01Mother fair or poor health −0.12 0.16∗ −0.05 0.04† 0.10 0.04 0.14 0.04†Father fair or poor health −0.01 0.10 −0.10 0.05∗∗ 0.01 0.03 −0.03 0.02Mother substance abuse 0.01 0.12 −0.05 0.05† 0.07 0.03† 0.08 0.03†Father substance abuse −0.03 0.08 0.01 0.03 0.01 0.02 0.03 0.02Mother reports domestic violence 0.01 0.18 −0.03 0.11 0.12 0.10 0.09 0.06Father impulsivity −0.03 0.05 −0.11 0.02∗∗ 0.03 0.01 0.01 0.01Father prior incarceration −0.04 0.08 −0.02 0.03 0.12 0.02∗∗ 0.11 0.01∗∗Dependent variable (lagged) 0.39 0.04∗∗∗ 0.39 0.04∗∗∗ 0.34 0.05∗∗∗ 0.09 0.06Constant 3.11 1.97 0.26 0.81R2 .30 .27 .49 .17

    Note: Data are from the Fragile Families and Child Wellbeing Study. Coefficients are standardized; robust standard errorsare listed as well. All analyses are weighted by the inverse probability of treatment (i.e., the probability of being in a relationshipbetween the 3- and 5-year surveys).

    aReference category: non-Hispanic White. bReference category: married. cReference category: less than high schooldiploma.

    †p< .10. ∗p< .05. ∗∗p< .01. ∗∗∗p< .001 (two-tailed tests).

    reference group was couples in which thefather had been previously but not recentlyincarcerated. I do not present these additionalanalyses because results were nearly identicalto those presented.

    The nonindependence of the couples wasanother threat to estimating nonbiased results,and, in supplemental analyses, I implementedseemingly unrelated regressions, which al-lowed for the estimation of two equations

  • 488 Journal of Marriage and Family

    Table 4. Ordinary Least Squares Regression Models Estimating Father-Reported Relationship Quality (N= 1,585)

    Overallrelationship quality Supportiveness Emotional abuse Physical abuse

    Variable 𝛽 SE 𝛽 SE 𝛽 SE 𝛽 SE

    Father recent incarceration −0.09 0.17 0.00 0.07 0.05 0.05 0.13 0.04∗

    Father current incarceration 0.16 0.21∗∗∗ 0.08 0.14 −0.06 0.07 −0.08 0.06†

    Constant 3.91 2.07 0.31 0.17R2 .21 .19 .40 .73

    Note: Data are from the Fragile Families and Child Wellbeing Study. Coefficients are standardized; robust standard errorsare listed as well. All analyses are weighted by the inverse probability of treatment (i.e., the probability of being in a relationshipbetween the 3- and 5-year surveys).

    †p< .10. ∗p< .05. ∗∗∗p< .001 (two-tailed tests).

    simultaneously and took into account the cor-relation of errors across the two equations(Greene, 2002). The results from these seem-ingly unrelated regression models producedcoefficients and standard errors that were almostidentical to those from OLS models and there-fore are not presented here. Because individualswere clustered in cities, all analyses includedrobust standard errors.

    Results

    Bivariate Association Between Incarcerationand Relationship Quality

    Table 2, which presents descriptive statistics forthe four measures of relationship quality sepa-rately for couples who did and did not include arecently incarcerated father, showed substantialvariation by incarceration. Among couples in arelationship at the 5-year survey, recent paternalincarceration was associated with lower relation-ship quality among both parents. For example,mothers in relationships with recently incarcer-ated men reported an overall relationship qualityof 3.35. This is in comparison to their counter-parts in relationships with not recently incarcer-ated men, who reported an overall relationshipquality of 3.99. Mothers in relationships withrecently incarcerated fathers also reported lesssupportiveness (2.50 compared to 2.62), moreemotional abuse (1.20 compared to 1.12), andmore physical abuse (1.12 compared to 1.03).These differences, all of which were statisticallysignificant, were medium to large in magnitude(Cohen’s d = .66 for overall relationship qual-ity, .32 for supportiveness, −.38 for emotionalabuse, and −.56 for physical abuse).

    The patterns of fathers’ reports of relationshipquality by recent paternal incarceration mirroredthe patterns of mothers’ reports. Recently incar-cerated fathers, compared to their counterparts,reported lower overall relationship quality (3.82compared to 4.17), less supportiveness (2.60compared to 2.70), more emotional abuse (1.26compared to 1.16), and more physical abuse(1.18, compared to 1.07). Again, these differ-ences were statistically significant and mediumto large in magnitude (Cohen’s d = .40 for over-all relationship quality, .28 for supportiveness,−.45 for emotional abuse, and− .53 for physicalabuse).

    Estimating Mothers’ and Fathers’ RelationshipQuality

    Given the selectivity of recently incarceratedmen, as described above, it is possible that theabove associations result not from incarcerationbut from other individual-level characteristicsassociated with incarceration and relation-ship quality. The data in Table 3 consider themultivariate association between paternal incar-ceration and mothers’ relationship quality.These models, which adjusted for a wide arrayof covariates, including a lagged dependentvariable, showed that recent paternal incarcer-ation was associated with overall relationshipquality (𝛽 =−.14, p< .01) and supportiveness(𝛽 =−.15, p< .05) and was marginally associ-ated with physical abuse (𝛽 = .12, p< .10). Butthe statistically significant bivariate associationbetween recent paternal incarceration and emo-tional abuse, observed in Table 2, was no longerstatistically significant when I controlled forfactors that were associated with selection into

  • Incarceration and Relationship Quality 489

    incarceration (though the relationship was in theexpected direction, 𝛽 = .02, ns).

    Table 3 also shows that, for mothers, currentpaternal incarceration operated differently thanrecent paternal incarceration. Mothers in rela-tionships with currently incarcerated fathers,compared to mothers not in relationships withcurrently incarcerated fathers, reported moresupportiveness (𝛽 = .11, p< .05). In addition,current paternal incarceration was negativelyassociated with physical abuse (𝛽 =−.12,p< .01), consistent with expectations, giventhat these couples had few, if any, opportunitiesto engage in physical abuse. It is important tonote that the coefficients for recent and currentpaternal incarceration were statistically differentfrom one another (p= .010 for supportivenessand p= .017 for physical abuse).

    In addition, the control variables operatedin the expected direction and magnitude. Forexample, the estimates for the full sampleshowed that non-Hispanic Blacks, comparedto non-Hispanic Whites, reported lower over-all relationship quality (𝛽 =−.09, p< .05).Cohabiting mothers reported lower relation-ship quality than married mothers (𝛽 =−.07,p< .10), and mothers’ health (𝛽 =−.12, p< .05)was inversely associated with overall relation-ship quality. Importantly, for estimates of bothoverall relationship quality and supportiveness,only one variable was larger in magnitude thanrecent or current paternal incarceration: thelagged dependent variable. Prior overall rela-tionship quality was positively associated withoverall relationship quality at the 5-year survey(𝛽 = .39, p< .001), and prior supportiveness waspositively associated with supportiveness at the5-year survey (𝛽 = .39, p< .001).

    Results estimating father-reported relation-ship quality are presented in Table 4. Thesemodels show that recently incarcerated fathers,compared to their not recently incarceratedcounterparts, reported more physical abuse(𝛽 = .13, p< .05). In these multivariate anal-yses, however, recent paternal incarcerationwas not associated with overall relationshipquality (𝛽 =−.09, ns), supportiveness (𝛽 = .00,ns), or emotional abuse (𝛽 = .05, ns). Currentincarceration was associated with higher overallrelationship quality (𝛽 = .16, p< .001) and wasmarginally associated with less physical abuse(𝛽 =−.08, p< .10). Again, the coefficients forrecent and current paternal incarceration werestatistically different from one another (p= .002

    for overall relationship quality and p= .026 forphysical abuse). The covariates, not presentedin the interest of parsimony, also operated in theexpected direction.

    Supplemental analyses

    Although the measure of recent paternalincarceration was appropriate and preciselymeasured, it is limited because it did not con-sider the possibility that the association betweenincarceration and relationship quality may varyby incarceration duration or offense type. Whenfathers experienced recent incarceration, moth-ers (but not fathers) were asked how long theincarceration lasted (or, if it was ongoing, itslength until the interview date) and the reasonfor the incarceration. In supplemental analyses,I first substituted the measure of recent paternalincarceration with dummy variables captur-ing incarceration duration: less than 3 months(2.6%), 3 months or longer (2.3%), missing(0.4%), and no recent incarceration (94.6%). Ithen substituted the measure of recent paternalincarceration with dummy variables captur-ing incarceration offense type: violent offense(0.7%), nonviolent offense (3.4%), missing(1.3%), and no recent incarceration (94.6%).The results provided virtually no evidencethat duration differentially affected relationshipquality or that offense type differentially affectedrelationship quality. There is one exception:When fathers were incarcerated for less than 3months, compared to when they were incarcer-ated for 3 months or longer, mothers reportedmore physical abuse (see Appendix Table A2).Given the few individuals in each of these cellsas well as the fact that many of these fathers hadbeen previously incarcerated (and, therefore, thismeasure does not represent lifetime duration),these results should be interpreted cautiously.

    Discussion

    This article extends the growing literature onthe collateral consequences of incarceration forfamily life by providing the first quantitativeexamination of the association between incar-ceration and relationship quality. Data fromthe FFCWB, a longitudinal survey of parents,documented two main findings that highlightthe complicated spillover effects of incarcer-ation. First, by and large, although paternalincarceration in the past 2 years was mostly

  • 490 Journal of Marriage and Family

    inconsequential for fathers’ reports of rela-tionship quality, mothers connected to theserecently incarcerated men reported lower over-all relationship quality, lower supportiveness,and greater physical abuse. Second, as currentpaternal incarceration is positively associatedwith some indicators of relationship quality, cur-rent paternal incarceration and recent paternalincarceration were differentially consequentialfor couples.

    Family stress theory, often used to describehow economic circumstances can impair rela-tionship quality, provides some guidance as towhy recent paternal incarceration is negativelyassociated with relationship quality among theromantic partners of the incarcerated. Incar-ceration is a stressor to the family system and,even when it occurs for only short periods oftime, is disruptive to family life. Maintaininga romantic relationship during incarcerationis economically, emotionally, and logisticallycomplicated, and the challenges associated withincarceration of a partner—for example, stigma,economic insecurity, or depression—do notsubside after release and, in some cases, may bemagnified while the formerly incarcerated tran-sition back to their pre-incarceration lives. Thesechallenges, along with the incapacitation andseparation associated with incarceration, likelymake maintaining high-quality relationshipsduring reintegration challenging.

    Recent paternal incarceration was not uni-formly associated with all four measures ofmother-reported relationship quality, given thatits effects on emotional abuse likely result fromselection processes. It seems reasonable toassume that positive and negative aspects ofrelationships are inversely correlated and thatincarceration would be similarly associated withsupportiveness because it is associated withemotional and physical abuse. But althoughpositive aspects of relationship quality are con-sidered less often than negative aspects (White& Rogers, 2000), these findings are consistentwith other research showing that the predictorsof positive aspects of relationship quality arenot necessarily the same as the predictors ofnegative aspects of relationship quality. Forexample, one examination of the link betweeneconomic factors and relationship quality foundthat economic factors predict conflict, but notaffection (Hardie & Lucas, 2010).

    Although family stress theory provides aframework for understanding the association

    between paternal incarceration and mothers’reports of relationship quality, it provides littleguidance as to why recent paternal incarcerationis, with one exception, not associated withfathers’ reports of relationship quality. Severalpossibilities may explain why paternal incar-ceration differentially affects the incarceratedand their romantic partners. One possibilityis that incarceration alters men’s personalities(e.g., socializing them to become violent) and,upon release, romantic partners (but not themen themselves) have a difficult time adjustingto these altered personalities. Another relatedpossibility is that fathers, happy to no longerbe behind prison walls, are simply less likelyto notice relationship difficulties. Although itis beyond the scope of this article—and thesedata—to adjudicate between these and otherpossibilities, doing so is an important direc-tion for future research. More generally, futureresearch should investigate the mechanismslinking incarceration to relationship quality.These include stigma and the resentment thataccompanies stigma (Braman, 2004), fear thata partner will be sent back to prison or jail(Goffman, 2009), fathers’ identity transforma-tion (Arditti et al., 2005), the dashed hopesfollowing optimism during incarceration (Bra-man, 2004; Roy & Dyson, 2005), and the ideathat incarceration increases the perception “rela-tionships are inherently exploitative” (Braman,2004, p. 88).

    Furthermore, the findings suggest thatalthough recent paternal incarceration is dele-terious to some measures of mothers’ (overallrelationship quality, supportiveness, physicalabuse) and fathers’ (physical abuse) relation-ship quality, current paternal incarceration isbeneficial for some measures of mothers’ (sup-portiveness, physical abuse) and fathers’ (overallrelationship quality, physical abuse) relation-ship quality. These findings help reconcilethe differences between quantitative researchsuggesting that incarceration destroys romanticrelationships by leading to dissolution (e.g.,Apel et al., 2010; Massoglia et al., 2011; West-ern, 2006) and qualitative research describingincarceration as a time when couples becomeoptimistic about and recommitted to their rela-tionships (Braman, 2004; Comfort, 2008). Thisethnographic research often examines couplesduring one partner’s incarceration and rarelyfollows couples into their postincarcerationlives (though see Braman, 2004). The fact

  • Incarceration and Relationship Quality 491

    that current and recent paternal incarcera-tion have countervailing consequences forrelationship quality suggests that future researchmust continue to rigorously interrogate thetiming of incarceration’s effects on family life.

    Limitations

    It is important to note that these analyseshave several limitations. First, the sample wasrestricted to couples in relationships at the 3-and 5-year surveys. Importantly, though, theanalyses accounted for selection into roman-tic relationships, which is in contrast to mostresearch examining predictors of relationshipquality. On a related note, I observed these rela-tionships during a relatively short time windowand, given that poor relationship quality predictsrelationship dissolution (Gottman, 1994), theseunions may eventually dissolve. Furthermore,these analyses are limited by threats to causalinference. Although it is infeasible to randomlyassign men to incarceration, and the analysespaid careful attention to causal processes (e.g.,considering the time ordering of variables, con-trolling for a wide array of characteristics), theobserved associations should not be interpretedas causal.

    There are additional limitations regardingvariable measurement. To begin, the incarcera-tion measure did not distinguish between prisonsand jails despite the fact that these two types ofcorrectional facilities may differentially affectrelationship quality. Prison incarceration mayhave more detrimental effects than jail incarcer-ation, because prisons are often located far fromromantic partners, and future survey designsshould collect information about correctionalfacilities. In addition, the alpha coefficients forseveral of the control variables are low, whichmay signal that the individual items are notmeasuring the same construct (though, withthe exception of pro-marriage attitudes, factoranalysis suggests this is not true). I includedthese variables in the analyses despite their lowalphas because they are theoretical predictorsof relationship quality and have been usedextensively in prior research (e.g., Tach & Edin,2013). Finally, paternal incarceration was cap-tured by both mothers’ and fathers’ reports, andboth parents reported on relationship quality,suggesting that shared methods variance maybias the results.

    Conclusion

    Taken together, these results suggest that cyclingthrough the criminal justice system, which hasbecome increasingly common among disadvan-taged men in the United States, has deleteriousconsequences for the quality of relationshipsbetween parents. Some couples remain inrelationships during and after incarceration,perhaps because of the anticipation and opti-mism associated with the incarceration period,but ultimately these relationships do not thrive(compared to their starting point or comparedto couples who do not experience paternalincarceration). These nuanced findings, whichare consistent with other research documentingthe consequences of incarceration for familylife (e.g., Turney & Wildeman, 2013), point toa neglected yet important piece of the incarcer-ation ledger (Sampson, 2011). It is now wellknown that researchers need to understand thespillover effects of incarceration on family lifeto construct an appropriate incarceration ledger.However, to not underestimate the consequencesof incarceration for family life, researchers mustalso consider the quality of relationships amongcouples who remain together.

    Note

    This publication was supported by grant number AE00102from the U.S. Department of Health and Human Services,Office of the Assistant Secretary for Planning and Evalua-tion (ASPE), which was awarded by the Substance Abuseand Mental Health Services Administration (SAMHSA).Its contents are solely the responsibility of the authorand do not necessarily represent the official views ofASPE or SAMHSA. Funding for the Fragile Familiesand Child Wellbeing Study was provided by the NationalInstitute of Child Health and Human Development throughGrants R01HD36916, R01HD39135, and R01HD40421,as well as a consortium of private foundations (see http://www.fragilefamilies.princeton.edu/funders.asp for thecomplete list). I am grateful to Jessica Hardie and AnitaZuberi for helpful comments on this manuscript.

    References

    Amato, P. R., & Rogers, S. J. (1999). Do attitudestoward divorce affect marital quality? Journal ofFamily Issues, 20, 69–86. doi:10.1177/019251399020001004

    Apel, R., Blokland, A. A. J., Nieuwbeerta, P., &van Schellen, M. (2010). The impact of imprison-ment on marriage and divorce: A risk set matchingapproach. Journal of Quantitative Criminology,26, 269–300. doi:10.1007/s10940-009-9087-5

  • 492 Journal of Marriage and Family

    Arditti, J. A., Smock, S. S., & Parkman, T. (2005).“It’s been hard to be a father”: A qualitative explo-ration of incarcerated fatherhood. Fathering, 3,267–288. doi:10.3149/fth.0303.267

    Booth, A., & Johnson, D. R. (1994). Declining healthand marital quality. Journal of Marriage and theFamily, 56, 218–223.

    Bradbury, T. N., Fincham, F. D., & Beach, S. R. H.(2000). Research on the nature and determinants ofmarital satisfaction: A decade in review. Jour-nal of Marriage and the Family, 62, 964–980.doi:10.1111/j.1741-3737.2000.00964.x

    Braman, D. (2004). Doing time on the outside: Incar-ceration and family life in urban America. AnnArbor: University of Michigan Press.

    Brown, S. L., & Booth, A. (1996). Cohabitation versusmarriage: A comparison of relationship quality.Journal of Marriage and the Family, 58, 668–678.

    Comfort, M. (2008). Doing time together: Love andfamily in the shadow of the prison. Chicago: Uni-versity of Chicago Press.

    Conger, R. D., Elder, G. H., Jr., Lorenz, F. O., Conger,K. J., Simons, R. L., Whitbeck, L. B., . . . Melby,J. N. (1990). Linking economic hardship to maritalquality and instability. Journal of Marriage and theFamily, 52, 643–656.

    Crosnoe, R., & Cavanagh, S. E. (2010). Familieswith children and adolescents: A review, critique,and future agenda. Journal of Marriage and Fam-ily, 72, 594–611. doi:10.1111/j.1741-3737.2010.00720.x

    Edin, K., Nelson, T. J., & Paranal, R. (2004). Father-hood and incarceration as potential turning pointsin the criminal careers of unskilled men. In M.Patillo, D. Weiman, & B. Western (Eds.), Impris-oning America: The social effects of mass incarcer-ation (pp. 46–75). New York: Russell Sage Foun-dation.

    Geller, A. (2013). Paternal incarceration andfather–child contact in Fragile Families. Journalof Marriage and Family, 75, 1288–1303. doi:10.1111/jomf.12056

    Geller, A., Cooper, C. E., Garfinkel, I., Schwartz-Soicher, O., & Mincy, R. B. (2012). Beyond absen-teeism: Father incarceration and its effects onchildren’s development. Demography, 49, 49–76.doi:10.1007/s13524-011-0081-9

    Glaze, L. E. (2011). Correctional populations inthe United States, 2010. Washington, DC: U.S.Department of Justice.

    Glenn, N. (1990). Quantitative research on maritalquality in the 1980s: A critical review. Journal ofMarriage and the Family, 52, 818–831.

    Goffman, A. (2009). On the run: Wanted men ina Philadelphia ghetto. American SociologicalReview, 74, 339–357. doi:10.1177/000312240907400301

    Gottman, J. M. (1994). What predicts divorce? Therelationship between marital processes and mari-tal outcomes. New York: Routledge.

    Greene, W. H. (2002). Econometric analysis (5th ed.).Englewood Cliffs, NJ: Prentice Hall.

    Grych, J. H., & Fincham, F. D. (1990). Marital conflictand children’s adjustment: A cognitive-contextualframework. Psychological Bulletin, 108, 267–290.doi:10.1037/0033-2909.108.2.267

    Hardie, J. H., & Lucas, A. (2010). Economic factorsand relationship quality among young couples:Comparing cohabitation and marriage. Jour-nal of Marriage and Family, 72, 1141–1154.doi:10.1111/j.1741-3737.2010.00755.x

    House, J. S., Landis, K. R., & Umberson, D. (1988,July 29). Social relationships and health. Science,241, 540–545.

    Karney, B. R., & Bradbury, T. N. (1995). The lon-gitudinal course of marital quality and stability:A review of theory, methods, and research. Psy-chological Bulletin, 118, 3–34. doi:10.1037/0033-2909.118.1.3

    Kurdek, L. A. (1989). Relationship quality for newlymarried husbands and wives: Marital history,stepchildren, and individual-difference predic-tors. Journal of Marriage and the Family, 51,1053–1064.

    Lavee, Y., McCubbin, H. I., & Olson, D. H. (1987).The effect of stressful life events and transitionson family functioning and well-being. Journal ofMarriage and the Family, 49, 857–873.

    Lopoo, L. M., & Western, B. (2005). Incarcerationand the formation and stability of marital unions.Journal of Marriage and Family, 67, 721–734.doi:10.1111/j.1741-3737.2005.00165.x

    Lowenstein, A. (1984). Coping with stress: The caseof prisoners’ wives. Journal of Marriage and theFamily, 46, 699–708.

    Massoglia, M., Remster, B., & King, R. D.(2011). Stigma or separation? Understanding theincarceration–divorce relationship. Social Forces,90, 133–155. doi:10.1093/sf/90.1.133

    McCubbin, H. I. (1979). Integrating coping behaviorin family stress theory. Journal of Marriage andthe Family, 41, 237–244.

    Nurse, A. M. (2002). Fatherhood arrested: Parentingfrom within the juvenile justice system. Nashville,TN: Vanderbilt University Press.

    Patterson, J. M. (2002). Integrating family resilienceand family stress theory. Journal of Marriage andFamily, 64, 349–360. doi:10.1111/j.1741-3737.2002.00349.x

    Pearlin, L. (1989). The sociological study of stress.Journal of Health and Social Behavior, 30,241–256.

    Reichman, N. E., Teitler, J. O., Garfinkel, I., &McLanahan, S. S. (2001). Fragile Families: Sam-ple and design. Children and Youth Services

  • Incarceration and Relationship Quality 493

    Review, 23, 303–326. doi:10.1016/S0190-7409(01)00141-4

    Rindfuss, R. R., & Stephen, E. H. (1990). Maritalnoncohabitation: Separation does not make theheart grow fonder. Journal of Marriage and theFamily, 52, 259–270.

    Roy, K. M., & Dyson, O. L. (2005). Gatekeeping incontext: Babymama drama and the involvementof incarcerated fathers. Fathering, 3, 289–310.doi:10.3149/fth.0303.289

    Sampson, R. J. (2011). The incarceration ledger:Toward a new era in assessing societal con-sequences. Criminology & Public Policy, 10,819–828. doi:10.1111/j.1745-9133.2011.00756.x

    Schnittker, J., Massoglia, M., & Uggen, C. (2012).Out and down: Incarceration and psychiatric dis-orders. Journal of Health and Social Behavior, 53,448–464. doi:10.1177/0022146512453928

    Schwartz-Soicher, O., Geller, A., & Garfinkel, I.(2011). The effect of paternal incarceration onmaterial hardship. Social Service Review, 85,447–473.

    Tach, L., & Edin, K. (2013). The compositionaland institutional sources of union dissolutionfor married and unmarried parents in the UnitedStates. Demography, 50, 1789–1818. doi:10.1007/s13524-013-0203-7

    Turney, K., Schnittker, J., & Wildeman, C. (2012).Those they leave behind: Paternal incarcerationand maternal instrumental support. Journal ofMarriage and Family, 74, 1149–1165. doi:10.1111/j.1741-3737.2012.00998.x

    Turney, K., & Wildeman, C. (2013). Redefiningrelationships: Explaining the countervailing con-sequences of paternal incarceration for parenting.American Sociological Review, 78, 949–979.doi:10.1111/j.1741-3737.2012.00998.x

    Wakefield, S., & Uggen, C. (2010). Incarcerationand stratification. Annual Review of Sociology,36, 387–406. doi:10.1146/annurev.soc.012809.102551

    Waller, M. R., & McLanahan, S. S. (2005). “His”and “her” marriage expectations: Determinantsand consequences. Journal of Marriage and Fam-ily, 67, 53–67. doi:10.1111/j.0022-2445.2005.00005.x

    Webster, P. S., Orbuch, T. L., & House, J. S. (1995).Effects of childhood family background on adultmarital quality and perceived stability. AmericanJournal of Sociology, 101, 404–432.

    Western, B. (2006). Punishment and inequality inAmerica. New York: Russell Sage Foundation.

    Western, B., Lopoo, L. M., & McLanahan, S. (2004).Incarceration and the bonds between parents inFragile Families. In M. Patillo, D. Weiman, & B.Western (Eds.), Imprisoning America: The socialeffects of mass incarceration (pp. 21–45). NewYork: Russell Sage Foundation.

    White, L., & Rogers, S. J. (2000). Economic cir-cumstances and family outcomes: A review ofthe 1990s. Journal of Marriage and the Family,62, 1035–1051. doi:10.1111/j.1741-3737.2000.01035.x

    Wildeman, C., & Muller, C. (2012). Mass impris-onment and inequality in health and family life.Annual Review of Law and Social Science, 8,11–30. doi:10.1146/annurev-lawsocsci-102510-105459

    Wildeman, C., Schnittker, J., & Turney, K. (2012).Despair by association? The mental health ofmothers with children by recently incarceratedfathers. American Sociological Review, 77,216–244. doi:10.1177/0003122411436234

  • 494 Journal of Marriage and Family

    APPENDIXTable A1. Logistic Regression Model Estimating the Propensity for Being Included in the Analytic Sample (N= 3,841)

    Variable OR SE

    Race/ethnicitya

    Non-Hispanic Black 0.57 0.11∗∗∗

    Hispanic 0.98 0.13Non-Hispanic other race 1.04 0.26

    Mother and father mixed-race couple 0.63 0.12∗∗∗

    Mother immigrant 2.04 0.14∗∗∗

    Mother age 1.02 0.01∗∗

    Mother lived with both parents at 15 1.21 0.08∗

    Mother and father relationship statusb

    Cohabiting 0.24 0.14∗∗∗

    Nonresidential romantic 0.14 0.14∗∗∗

    No relationship 0.08 0.19∗∗∗

    Relationship duration (years) 1.02 0.01∗

    Relationship commitment 1.92 0.07∗∗∗

    Mother pro-marriage attitudes 1.18 0.11Mother traditional attitudes 1.10 0.07Mother gender distrust 0.86 0.08†

    Mother educationc

    High school diploma or GED 0.98 0.10More than high school 1.07 0.11

    Mother employment 0.98 0.09Mother income-to-poverty ratio 1.06 0.02∗

    Father prior incarceration 0.75 0.09∗∗

    Constant −1.13Log likelihood −2,006

    Note: Data are from the Fragile Families and Child Wellbeing Study. Robust standard errors are listed. OR= odds ratio.aReference category: non-Hispanic White. bReference category: married. cReference category: less than high school

    diploma.†p< .10. ∗p< .05. ∗∗p< .01. ∗∗∗p< .001 (two-tailed tests).

  • Incarceration and Relationship Quality 495

    Table A2. Ordinary Least Squares Regression Models Estimating Mother-Reported Relationship Quality (N= 1,848)

    Overallrelationship quality Supportiveness Emotional abuse Physical abuse

    Variable 𝛽 SE 𝛽 SE 𝛽 SE 𝛽 SE

    Estimating mother-reported outcomesIncarceration durationa

    Less than 3 months −0.13 0.22∗∗ −0.08 0.08∗ 0.06 0.04 0.14 0.04∗∗

    3 months or greater −0.05 0.23 −0.10 1.04† −0.01 0.05 0.00 0.04Missing −0.03 0.31 −0.14 0.20∗ −0.11 0.09† −0.05 0.04Statistically different coefficients? No No No Yes

    Incarceration offense typea

    Violent offense −0.03 0.33 −0.02 0.11 0.02 0.07 0.11 0.11Nonviolent offense −0.14 0.21∗∗ −0.10 0.08∗ 0.03 0.04 0.10 0.03∗

    Missing −0.05 0.29 −0.14 0.13∗ −0.03 0.07 0.02 0.05Statistically different coefficients? No No No No

    Estimating father-reported outcomesIncarceration durationa

    Incarceration for less than 3 months −0.07 0.19 0.02 0.10 0.02 0.06 0.08 0.05†

    Incarceration for 3 months or greater −0.09 0.24 −0.05 0.07 0.07 0.08 0.09 0.05†

    Missing 0.07 0.22∗ 0.03 0.10 −0.01 0.06 0.05 0.08Statistically different coefficients? No No No No

    Incarceration offense typea

    Violent offense −0.02 0.26 0.00 0.14 0.05 0.13 0.07 0.10Nonviolent offense −0.07 0.16 −0.02 0.09 0.00 0.04 0.06 0.04Missing −0.06 0.33 0.03 0.09 0.09 0.10 0.13 0.06∗

    Statistically different coefficients? No No No No

    Note: Data are from the Fragile Families and Child Wellbeing Study. Coefficients are standardized; robust standard errorsare listed as well. Ns for father-reported relationship quality are smaller (N = 1,585). All analyses are weighted by the inverseprobability of treatment (i.e., the probability of being in a relationship between the 3- and 5-year surveys). Rows indicating“Statistically different coefficients?” refer to difference in coefficients between less than 3 months and 3 months or greater (forincarceration duration) and violent offense and nonviolent offense (for incarceration offense type).

    aReference category: no incarceration.†p< .10. ∗p< .05. ∗∗p< .01. ∗∗∗p< .001 (two-tailed tests).