Married and Cohabiting Parents’ Relationship Stability: A Focus on Race and Ethnicity

CYNTHIA OSBORNE University of Texas at Austin

WENDY D. MANNING Bowling Green State University*

PAMELA J. SMOCK The University of Michigan**

Married and Cohabiting Parents’ Relationship

Stability: A Focus on Race and Ethnicity

We draw on three waves of the Fragile FamiliesStudy (N ¼ 2,249) to examine family stabilityamong a recent birth cohort of children. Wefind that children born to cohabiting versusmarried parents have over five times the risk ofexperiencing their parents’ separation. This dif-ference in union stability is greatest for Whitechildren, as compared with Black or MexicanAmerican children. For White children, differ-ences in parents’ education levels, paternalsubstance abuse, and prior marriage and chil-dren account for the higher instability faced bythose born to cohabiting parents, whereas dif-ferences in union stability are not fully ex-plained among Black and Mexican Americanchildren. These findings have implications forpolicies aimed at promoting family stability andreducing inequality.

The rise in cohabitation is well documented, withcohabitation playing an increasingly prominent

role in the lives of American adults and children.At least one child in eight is born into a cohabitingparent family, and this rate has doubled over thecourse of a single decade (Bumpass & Lu,2000). Thus, not surprisingly, researchers havebegun to focus on the implications of cohabita-tion for children’s well-being (Brown, 2004a;Manning & Lamb, 2003; Osborne, McLanahan,& Brooks-Gunn, 2005).

A primary reason why parental cohabitationis expected to influence children’s well-being isthat cohabiting unions are typically short lived(Bumpass & Lu, 2000), and family stabilityhas consistently been associated with positivechild well-being (Amato, 2005; McLanahan &Sandefur, 1994; Osborne & McLanahan, 2007).Indeed, one recent study showed that children born10 – 25 years ago to cohabiting parents face a sig-nificantly higher risk of experiencing family in-stability than children born to married parents(Manning, Smock, & Majumdar, 2004).

We use three waves of data from the FragileFamilies and Child Wellbeing Study (FragileFamilies Study) to examine the early life courseof a recent birth cohort (1998 – 2000) of childrenwho were born into cohabiting unions, contrast-ing the stability of their parents’ unions to thoseof children born in marriages. Focusing ona recent birth cohort is important because chil-dren are more likely to be born into cohabitingunions today (Casper & Bianchi, 2002; Fields,2004) and cohabiting unions have becomeincreasingly less stable (Bumpass & Lu, 2000).

LBJ School of Public Affairs, Population Research Center,University of Texas at Austin, 2315 Red River Road, P.O.Box Y, Austin, TX 78713 ([email protected]).

*Department of Sociology, Center for Family and Demo-graphic Research, 233 Williams Hall, Bowling Green StateUniversity, Bowling Green, OH 43403.

**Department of Sociology, Population Studies Center, TheUniversity of Michigan, 426 Thompson Street, Ann Arbor,MI 48106.

Key Words: child well-being, cohabitation, family demogra-phy, family structure, marriage, union dissolution.

Journal of Marriage and Family 69 (December 2007): 1345–1366 1345

Our study examines the extent to which beingborn to cohabiting parents increases the likeli-hood of experiencing the end of the parents’union. Moreover, we examine why union statusdifferentials in family stability may exist. In thisaim, we move beyond prior work by includinga more comprehensive array of potential mediat-ing variables to explain differential stability forchildren born to cohabiting versus married pa-rents (Manning et al., 2004). The prospectivestudy design allows us to evaluate the extent towhich the gap in relationship stability is associ-ated with fewer economic resources, familial sup-port, relationship quality, and family complexityincluding relationship and fertility histories.Throughout, we also focus on similarities and dif-ferences for Black, Mexican American, andWhite children because of evidence that theprominence and role of cohabitation in familyformation varies by race and ethnicity (Smock,2000). Children are much more likely to be pres-ent in minority cohabiting couple households(67% and 70% among Blacks and Hispanics,respectively) than in White cohabiting house-holds (35%) (McLanahan & Casper, 1995), andWhite children spend less of their childhood incohabiting parent families than Black or Hispanicchildren (Bumpass & Lu). Similarly, there areracial and ethnic differentials in the proportionof children being born to cohabiting parents.Among Whites, only about 1 in 10 children isnow born into cohabiting parent families com-pared with nearly 1 in 5 for Black and Hispanicchildren (Bumpass & Lu). Recent research indi-cates that some of the race and ethnic differencesin family formation may be related to beingforeign- versus U.S. born (Brown, Van Hook, &Glick, 2006; Phillips & Sweeney, 2005; Raley,Durden, & Wildsmith, 2004), so we also accountfor the nativity status of Mexican Americanmothers.

WHY COHABITATION MATTERS

Children born to married parents may be lesslikely to experience their parents’ separationbecause marriage may confer unique economicadvantages (Becker, 1991), emotional benefits(Waite, 2000), assistance from extended family(Eggebeen, 2005), and institutional support(Cherlin, 2004) to the parents that, in turn, pro-mote greater union stability. Cohabitation is sim-ilar to marriage in many respects; two adults sharean intimate relationship and a household, and in

our sample they also share a child. Thus, wemight expect the same benefits to be conferredto cohabiting parents. Yet, because cohabitationis considered to be a less committed relationshipand it is less institutionalized than marriage, itmay not award the same benefits as marriage interms of economic resources, familial support,and relationship quality, which may weaken thestability of the union. It is also likely that manyof the differences between married and cohabit-ing parents in factors associated with union sta-bility, such as sociodemographic factors andfamily complexity, actually predate the forma-tion of their union, rather than being caused bythe union itself.

With regard to economic resources, the empir-ical evidence shows that married parents haveconsiderably higher incomes and educationalattainment than cohabiting parents (Manning& Brown, 2006), and economic resources arestrongly and positively correlated with union sta-bility (Fein, Burstein, Fein, & Lindberg, 2003).The extent to which the link between marriageand higher levels of economic resources is causalrather than attributable to selection is less conclu-sive, but it is likely the result of both. Researchersusing twin studies (Antonovics & Town, 2004)and other advanced methods to account for un-observed heterogeneity between married andunmarried men (Ginther & Zavodny, 2001;Korenman & Neumark, 1997) argue that the mar-riage wage premium for men is largely causal. Co-habiting men also enjoy a wage premium ascompared with single men, yet this premium issmaller than married men’s (Cohen, 2002) and isprimarily the result of selection (Stratton, 2002).These studies are not specific to fathers, however.In addition, cohabitation may lead to the accumu-lation of fewer economic resources than marriagebecause cohabiting parents often do not share ex-penses or pool resources to the extent that marriedparents do (Kenney, 2003); insomuch as marriedcouples anticipate longer relationship stability,they may be more willing than cohabitors to makecooperative bargaining arrangements and poolhousehold resources and expenses (Lundberg &Pollack, 1994).

There is also evidence that marriage is selec-tive of couples with higher economic resourcesand this generalization holds when comparingmarried and cohabiting parents as well (Casper& Bianchi, 2002; Manning & Brown, 2006).Research focusing on marriage among cohabi-tors finds that cohabiting couples with greater

1346 Journal of Marriage and Family

economic resources and higher human capitalare more likely to marry following a premaritalconception (Manning, 1993, 2001) or birth(Osborne, 2005) rather than to remain cohabiting.In addition, many cohabiting parents desire mar-riage but forgo getting married until they canachieve financial stability and resolve many oftheir relationship problems (Gibson-Davis, Edin,& McLanahan, 2005; Manning & Smock, 2002;Smock, Manning, & Porter, 2005).

Another type of resource is familial or socialsupport. Social support, particularly from pa-rents, may bolster the couple’s relationship andenhance its stability, and it may be particularlyimportant for families with young children. Co-habiting couples are less able to access supportfrom family members and others (Brines &Joyner, 1999; Eggebeen, 2005), including finan-cial support (Hao, 1996) and instrumental sup-port (Marks & McLanahan, 1993), and they areless likely to report having expectations to turnto their parents in times of need (Eggebeen).These findings apply to cohabiting couples withchildren as well (Harknett & Knab, 2007). Lowerlevels of familial support directed toward cohab-iting parents may be related to issues surroundingnorms for childbearing outside of marriage,strained relationships with parents (Nock,1995), as well as strapped or complex family net-works (Harknett & Knab). Unlike prior studies,our data allow us to include measures of socialsupport, although we focus on perceived socialsupport rather than actual receipt of supportbecause receipt of support may be contingentupon need.

With regard to relationship quality, marriedcouples report greater relationship quality thancohabiting couples (Brown & Booth, 1996;DeMaris, 2000; Nock, 1995), and relationshipquality is positively correlated with relationshipstability (Carlson, McLanahan, & England,2004; Osborne, 2005; Roebuck Bulanda &Brown, 2007). Cohabiting couples have fewerbarriers to exit their relationship and thus maybe less likely to remain in a relationship with poorquality. There is no conclusive evidence, how-ever, that the link between marriage and relation-ship quality is causal (Brown, 2004b), andsubstantial evidence suggesting that marriage isselective of better relationships. Couples withgreater relationship happiness, less violence,and greater anticipated stability are more likelyto marry (Brown, 2000, 2004b; DeMaris, 2000),and these findings apply to unmarried parents as

well (Carlson et al.; Osborne). Previous workhas not considered whether relationship qualityexplains the gap in the stability of cohabitingand married parent families.

Compared with marriages, cohabiting unionscontain more complex families in terms of priorrelationships (marriage and cohabitation) andprior fertility (Carlson & Furstenberg, 2006;Goldscheider & Sassler, 2006; Manning, 2004;Stewart, Manning, & Smock, 2003), which isnotable because family complexity, in terms ofcohabitation and marital history, is associatedwith marital instability (Raley & Bumpass,2003; Sweeney & Phillips, 2004; Teachman,2004). Individuals who have a history of dissolv-ing marriages and cohabitations may be moreprone to end their current union. In our analyses,we account for relationship history and distin-guish between mothers’ marriage and cohabita-tion experience.

Parents with children from prior unions aremore likely to experience separation as comparedto parents with only biological children present(Carlson et al., 2004; Osborne, 2005; White &Booth, 1985) because relationships with childrenthat extend beyond household boundaries createsome stress in both the resident and nonresidentparents’ household. Cohabiting mothers and fa-thers are substantially more likely than marriedparents to have children from prior unions,whereas married parents are more likely to sharebiological children (Carlson & Furstenberg,2006; Osborne). To better understand whetherthe more complex nature of cohabiting relation-ships helps to explain their higher rates of separa-tion, we include measures of mothers’ andfathers’ fertility from prior unions as well asa measure of the timing of pregnancy of the focalchild relative to the marriage or cohabitation. Todate, prior studies have not accounted for rela-tionship history in terms of both cohabitationand marriage as well as parents’ fertility history.

RACE, ETHNICITY, AND NATIVITY STATUS

Our conceptual model incorporates many ele-ments tapping the potential benefits conferredby marriage, but we recognize that these benefits,as well as the selectivity of marriage versuscohabitation, may not be consistent across all raceand ethnic groups. Generally, differences in eco-nomic resources, social support, relationshipquality, and family complexity between marriedand cohabiting parents are expected to be smaller

Married and Cohabiting Parents’ Union Stability 1347

for Black and Mexican American parents as com-pared to White parents.

First, the gap in economic resources of cohab-iting and married parent households is smalleramong minority than White families (Manning,2001; Manning & Brown, 2006). Cohabitationappears to be more selective of the most disad-vantaged among Whites, more so than for Blacksor Mexican Americans (Bumpass & Lu, 2000;Raley & Wildsmith, 2004; Wildsmith & Raley,2006).

Second, although researchers have not exam-ined racial and ethnic differences in social sup-port in cohabitation and marriage, we predict itmay be greater among minority families. Consen-sual unions and childbearing within cohabitingunions are more normative among minority fam-ilies (Casper & Bianchi, 2002; Manning, 2001;Musick, 2002), and Black and Hispanic childrenspend more time in cohabiting parent unions thanWhite children (Bumpass & Lu, 2000).

Third, relationship quality is lower amongBlack married couples than White or MexicanAmerican married couples, but there is no racialgap in relationship quality among cohabitingcouples (Brown, 2003; Roebuck Bulanda &Brown, 2007). These results suggest that Blackcohabiting and married couples may have moresimilar relationship quality than White cohabit-ing and married couples, and thus, relationshipquality may explain more of the difference inunion stability for White than Black parents.

Fourth, family complexity is greater amongAfrican Americans because of higher levels ofpremarital childbearing, nonresident father-hood, multiple partner fertility, and divorce(Carlson & Furstenberg, 2006; Phillips &Sweeney, 2006). Although, there is a smallergap in family patterns among Whites and His-panics, teenage and premarital fertility is greateramong Hispanics than Whites (Hamilton, Mar-tin, & Sutton, 2003). Some of these indicators offamily complexity, such as premarital child-bearing, have been found to have a more desta-bilizing influence for White than Black marriedcouples (Sweeney & Phillips, 2004). Thus, weexpect that family complexity will have agreater effect on Whites than Blacks or MexicanAmericans.

Finally, marriages are less stable for Blackthan White families (Phillips & Sweeney, 2005;Raley & Bumpass, 2003; Teachman, 2004),whereas Mexican American and White marriedcouples share similar relationship stability

(Roebuck Bulanda & Brown, 2007). Thus, theunion status differentials in terms of cohabitationversus marriage may be smaller for Black parentfamilies as compared to Whites and MexicanAmericans.

Recent work has expanded on racial andethnic differentials in family formation behaviorby incorporating nativity status (Brown et al.,2006; Phillips & Sweeney, 2005; Raley et al.,2004). Native-born Mexican Americans moreoften cohabit than foreign-born Mexican Amer-icans, and the proportion married is loweramong native- than foreign-born Mexican Amer-icans (Brown et al.). Marital disruption amongnative-born Mexican Americans is similar to dis-ruption rates among Whites, whereas foreign-born Mexican Americans have significantlylower disruption rates than Whites (Phillips &Sweeney, 2005). Additionally, foreign-bornMexican Americans report fewer marital pro-blems than Whites or Blacks, but native-bornMexican Americans do not (Roebuck Bulanda& Brown, 2007). Several arguments have beenproposed to explain these differentials in familyformation behavior, including selection on thebasis of immigration, economic circumstances,the assimilation into American culture, and legalincentives to get and remain married. Althoughwe will not be able to determine the sourceof the differential, prior work has not consideredthe relative stability of cohabiting and marriedunions according to the nativity status ofparents.

Researchers theorize that cohabitation playsa different role in the family formation processacross race and ethnic groups; Blacks and Mexi-can Americans reportedly view cohabitation asmore of an alternative to marriage, whereasWhites consider cohabitation as a step towardmarriage or trial status for their relationship(Manning & Landale, 1996; Sweeney & Phillips,2004; Wildsmith & Raley, 2006). This depictionis consistent with the findings that the gap in indi-vidual (e.g., income or fertility) and relationshipcharacteristics (e.g., relationship quality andduration) is greater among White cohabiting andmarried couples than Black or Hispanic couples(Brown, 2003; Bumpass & Lu, 2000; Manning& Brown, 2006). Thus, we expect that the eco-nomic resources, familial support, relationshipquality, and family complexity measures mayexplain the gap in stability of cohabiting and mar-ital unions among Whites more so than Blacks orMexican Americans.


Moreover, prior research finds that the factorspredicting separation from marriage differ acrossrace and ethnic groups (Pagnini & Morgan, 1996;Phillips & Sweeney, 2005; Sweeney & Phillips,2004). To our knowledge, no research has deter-mined whether the factors predicting separationfrom cohabitation differs across groups, but it islikely. Thus, if the factors predicting separationdiffer, and the rates of separation differ, then itis likely that unique processes promote familystability among Whites, Blacks, and MexicanAmericans.

CURRENT INVESTIGATION

We have three primary goals: (a) to determine theextent to which children recently born to cohabit-ing parents face higher levels of parental insta-bility than their counterparts born to marriedparents; (b) to identify the factors that accountfor this association, paying particular attentionto the roles that economic resources, social sup-port, relationship quality, and family complexityplay in explaining the higher incidence of insta-bility among cohabiting parents; and (c) to exam-ine how these processes operate among specificracial and ethnic groups. Our study builds onextant knowledge in five central ways.

First, the analyses are based on a very recentbirth cohort of children. Prior studies are basedon much earlier birth cohorts, so our work pro-vides some of the most up to date analyses offamily stability. Only a few studies have directlycompared prospects for family stability for chil-dren born into cohabiting versus married couplefamilies. These include Landale and Hauan(1992), who study the family life courses ofPuerto Rican children born in the mid-1980s.They find that children born in cohabiting unionshave almost twice the odds of experiencing thebreakup of their parents’ unions (whether or notthe relationship was transformed into marriage)as children born in marriage, although the gapis narrowed with the inclusion of characteristicsof the mother, father, and the union. Researchon Canadian children indicates a similar pat-tern (Marcil-Gratton, Le Bourdais, & Lapierre-Adamcyk, 2000), as does descriptive researchon U.S. children born in the early 1980s (Raley& Wildsmith, 2004; Wu, Bumpass, & Musick,2001). Another study shows that children bornbetween 1980 and 1994 to cohabiting parentsmore often experience parental disruption thanchildren born to married parents, after controlling

for an array of characteristics of the mother(Manning et al., 2004).

Second, we use much more detailed measuresof socioeconomic status that are correlated withunion stability (Fein et al., 2003; Smock &Manning, 1997; Smock et al., 2005) and includethese measures for both mothers and fathers.Prior work is limited because it is restricted tocharacteristics of the mother and does not includean indicator of family income (Manning et al.,2004). Third, unlike prior studies, we includemeasures of relationship quality and familycomplexity that may influence union stability(Carlson et al., 2004; Osborne, 2005; Sweeney& Phillips, 2004).

Fourth, we include an extensive set of controlvariables that may better account for selectioninto either marriage or cohabitation for childbear-ing. Cohabitors are more likely than marriedparents to be younger, less religious, and to comefrom divorced homes, and each of these factors isnegatively associated with marriage and unionstability (Carlson et al., 2004; Osborne, 2005).Although we account for an extensive set of con-trol variables, we are unable to determine if themeasures included in this analysis are caused bythe parents’ relationship status or whether theyare predictive of their union status for childbear-ing. It is also likely that cohabiting and marriedparents differ in ways that we are unable to mea-sure using survey data, and thus, we may not beable to explain fully the differential stability oftheir unions.

Finally, our study pays close attention to racialand ethnic variation in the life courses of childrenborn to married versus cohabiting parents. Priorwork has not focused on nativity status, whichmay be important for understanding family sta-bility for Mexican American children. In addi-tion, prior studies have not thoroughly exploredhow family processes may differ across raceand ethnic groups.

METHOD

We use data from three waves of the Fragile Fam-ilies Study, a longitudinal birth cohort survey,which between 1998 and 2000 interviewed ap-proximately 3,500 unmarried mothers and 1,500married mothers in the hospital at their child’sbirth, in 20 large cities throughout the UnitedStates. About 87% of the mothers were reinter-viewed when the child was aged 1 (1999 – 2001)and 3 (2001 – 2003).


Our analyses are based on data from 2,249mothers who were either married (n ¼ 886;39%) or cohabiting (n ¼ 1,363; 61%) with theirchild’s biological father at the baby’s birth. Thissample excludes 1,928 mothers (39%) whowere not married to or living with their child’sbiological father at the child’s birth. An additional529 mothers (206 married and 323 cohabiting;11% of the original sample) who identified theirrace or ethnicity as something other than White,Black, or Hispanic of Mexican origin are alsoexcluded so that we may focus on differencesacross race or ethnic groups. Finally, an addi-tional 192 mothers (83 married and 109 cohabit-ing; 4% of the original sample) who did notcomplete the follow-up interviews are excluded.The characteristics of the mothers lost to follow-up are similar to the mothers who remain in thesample with the exceptions that excluded mothersare more likely to be Hispanic, have less educa-tion, and have fewer children with the biologicalfather or a prior partner as compared with moth-ers who remain in the sample.

Dependent Variable

Our dependent variable is the hazard of the sepa-ration of the couple’s relationship by the thirdwave of interviews. Separation is defined as thecouple no longer coresiding and is measured bythe month and year of separation as reported bythe mother in the second and third waves of thestudy. Mothers who report being separated atWaves 2 or 3 but who did not report a month oryear of separation (n ¼ 19; 0.8%) were codedas separating in January of the year of inter-view. We continue to consider a cohabitingrelationship as stable if the relationship transi-tions to marriage (n ¼ 391; 29% of cohabitors).Nine percent of cohabitors who married follow-ing their child’s birth (n ¼ 36) separated priorto the third interview date and are counted asseparating. Separation is based on a discreteevent and is subject to censoring at the thirdwave of interviews. Couples were not consid-ered at risk of separation until after the birth oftheir biological child.

Independent Variables

The mother’s self-reported relationship status(married or cohabiting) at the time of her child’sbirth is the main independent variable in this anal-ysis. At the child’s birth, mothers are considered

cohabiting if the mother reports that she lives withthe child’s father and is not married. Cohabitationis measured as living together most or all of thetime at the follow-up interviews. The results aresimilar if we limit our measurement of cohabita-tion to all of the time.

The other independent variables include themother’s race or ethnicity, duration of the unionprior to the child’s birth, mothers’ backgroundcharacteristics, mothers’ and fathers’ educationand economic resources, mothers’ perceivedsocial support, the couples’ relationship quality,and family complexity including mothers’ rela-tionship history and the parents’ fertility history.

We imputed missing data for the independentvariables to the mean of the subgroup (marriedor cohabiting). No data are missing for the pa-rents’ relationship status at birth; for the otherindependent variables, missing data representless than 2% of the cases on all variables exceptfor household income. For this variable, we usethe constructed Fragile Families household in-come variable that is generated using hot-deckingtechniques.

We use three dichotomous, self-identifiedmeasures for the mothers’ race or ethnicity(non-Hispanic White, non-Hispanic Black, andHispanic of Mexican origin). We exclude moth-ers who categorize their race as other or whocategorize their ethnicity as Hispanic but not ofMexican origin. Of the mothers in the samplewho identify as Hispanic, 61% are of Mexicanorigin. Because Hispanics of varying ethnicbackgrounds face quite different social and eco-nomic conditions, we limit our focus to onlyone group for which there is adequate samplesize. In addition, because immigrants face eco-nomic and social conditions that are often differ-ent from native-born Mexican Americans, wetake nativity status into consideration whenfocusing on Mexican Americans separately.

The duration of the parents’ union prior to thechild’s birth is based on the mothers’ report of themonth and year the couple began living togetherfor cohabiting parents and month and year ofmarriage for parents’ married at their child’sbirth. The questions are asked at the second-and third-wave interviews.

Controls for background characteristics of themother include age, family background, and reli-giosity. These variables are all measured at base-line. Mothers’ age is a continuous variable.Family background is a dichotomous measureindicating whether the mother’s parents were


married when she was 15 years old. Religiosity isa measure of frequency of attendance at a reli-gious service and is coded 1 ¼ yes, 0 ¼ no ifthe mother attends a religious service weekly.

Fathers’ education is based on four categories:less than high school, high school, some collegeor technical training, and college degree ormore. Because mothers’ and fathers’ education ishighly correlated (r ¼ .61), mothers’ educationis based on three categories, indicating that she isin a higher, lower, or similar education categoryas the father.

The economic characteristics of the parentsinclude annual household income and employ-ment. The measure of annual household incomeis linear, time varying, and is constructed fromthe mothers’ reports. Fathers’ and mothers’employment are also time-varying covariates.Time-varying covariates are measured at baselinefor couples who separate by Wave 2 and are mea-sured at Wave 2 for couples who are still togetherat Wave 2. Fathers’ employment is a dichotomousvariable measured as working for earnings in theweek prior to the baseline or the Wave 2 inter-view. Mothers’ employment is based on employ-ment from earnings in the year prior to the child’sbirth and in the week prior to the Wave 2 inter-view. These data do not provide informationregarding the stability or trajectory of themothers’ or fathers’ employment. Instability inemployment, however, should be reflected inannual income, which we do measure.

To measure perceived social support we usetwo dichotomous measures, indicating that some-one in the mother’s family could loan her $200 orhelp her with babysitting or child care. Thesemeasures are time varying. We include measuresof perceived support rather than received supportbecause receipt of support may be contingentupon need.

Our relationship quality measures have beenused widely and effectively in other studies andappear to be highly predictive of union transitions(Carlson et al., 2004; Osborne, 2005). We includefour characteristics of the couple’s relationshipquality: the emotional support the mother feelsfrom the father, the level of disagreement thecouple experienced in the month prior to thechild’s birth, maternal reports of domestic vio-lence, and substance abuse. The measure foremotional support the mother feels from thefather is a time-varying covariate and is basedon the mean of three questions including thebaby’s father is fair and willing to compromise,

expresses love and affection to the mother, andencourages the mother to do things important toher (a ¼ .69). The responses are recoded suchthat 3 is equal to often and 1 is equal to never.Disagreement within the relationship is mea-sured by a count of six variables from the moth-er’s report of disagreeing sometimes or oftenwith the father about money, spending timetogether, sex, the pregnancy, drugs or alcohol,and being faithful within the month prior totheir child’s birth. The measure of domestic vio-lence is dichotomous, time varying, and basedon the mothers’ report. At baseline, the measureis based on whether the father sometimes oroften hits or slaps the mother when he is angry;at the Wave 2 interview, the measure is basedon whether the father sometimes or often kicksor slaps her when he is angry. Mothers’ andfathers’ substance abuse measures are alsodichotomous and time varying. At baseline,mothers are considered to have a substanceabuse problem if they reported using drugs dur-ing their pregnancy. At Wave 2, mothers arecoded as having a substance abuse problem ifthey report smoking marijuana or using anyhard drug in the past month or if alcohol ordrugs have interfered with how they managetheir daily lives or their relationships since theirchild’s birth. For fathers, the mothers wereasked at baseline and Wave 2 if the father has adrug or alcohol problem that interferes with hiswork or relationships.

In addition, we control for several characteris-tics of the mothers and fathers to account for thecomplex nature of these families. These includemeasures of the mothers’ prior relationship insta-bility and the mothers’ and fathers’ fertility histo-ries. Fathers’ relationship history is not availablein the data. We operationalize prior relationshipinstability of the mother by dichotomous varia-bles indicating whether the mother was evermarried (and thus separated or divorced) or ina cohabiting relationship with another partnerprior to her current union. Although these ques-tions were asked at the third interview, our meas-ures refer to the period prior to the focal child’sbirth.

We also include measures of mothers’ and fa-thers’ fertility from prior unions as well as a mea-sure of the timing of pregnancy of the focal childand marriage or cohabitation. The parents’ fertil-ity history is measured by four dichotomous var-iables. One variable measures whether the coupleshared a biological child at the time of the current


child’s birth, which serves as an indicator of thecouple’s parity. We also include dichotomousmeasures for whether the mother has a child froma previous relationship or the father has a childfrom a previous relationship. A fourth variableis time varying and indicates whether the motherand father had additional children after the focalchild’s birth. The final indicator is the timing ofthe union formation relative to the pregnancy ofthe focal child. Couples who formed unions inresponse to a pregnancy may be less prepared tosustain a stable union, and the pregnancies maybe less often planned. The dichotomous measureindicates whether the focal child was born fewerthan 7 months following the formation of the pa-rents’ union status at the child’s birth.

Analytic Strategy

Our analyses include life-table estimates andevent history analyses. We use single-decrementlife-table estimates to estimate the hazard rate ofseparation or the cumulative proportion of mar-ried and cohabiting parents who experienceseparation within 3 years following the birth ofa child. We also use event history models to pre-dict the factors that influence the hazard that chil-dren born into cohabiting and married parentfamilies experience family instability by age 3.Specifically, we use Cox proportional hazardtechniques, which allow us to use time-varyingvariables and do not require us to assume a setprobability distribution (Allison, 1984).

After presenting descriptive statistics on ourindependent variables and discussing resultsfrom the life tables, we move to our multivariatemodels. We discuss the bivariate associationsbetween union status at birth and separation andpresent six models. The bivariate association isthe observed difference in the hazard of union sta-tus stability that we seek to explain. The firstmodel controls only for race and ethnicity. Thesecond model is our base model that controlsfor relationship duration prior to the child’s birth;background characteristics of the mother includ-ing race, age, family background, and religiosity;and the mothers’ and fathers’ education level. Wethen add in sets of variables to empirically evalu-ate how the inclusion of specific sets of covariatesinfluences the relationship between union statusand family stability. Covariates that are associ-ated with family instability and that differ in prev-alence between married and cohabiting parentswill help to explain the difference in family stabil-

ity between the two groups. In the third model, weadd in controls for mothers’ and fathers’ eco-nomic resources and perceived social support.In the fourth model, we remove the economic re-sources and support measures and control for thecouples’ relationship quality. In the fifth model,we control for measures of family complexityincluding relationship and fertility history. Thesixth model is the full model and includes allthe measures listed above.

To investigate racial and ethnic differences, weempirically evaluate how the sets of variables inModels 2 through 6 attenuate race or ethnic dif-ferences in addition to their association withunion status. We also test interactions betweenunion status at the child’s birth and race or ethnic-ity. On the basis of the results of the Chow test(p , .0001) (DeMaris, 2002) and the contrast ofthe log-likelihood for models with no interac-tions to models that include cross products of allcovariates with race and ethnicity (p , .0001),we present separate models for each race andethnic group. This fully interacted model alsoallows us to determine if the effect of certainvariables differs across race or ethnic groups. Inthe separate model for Mexican Americans, weinclude a variable for native born and an interac-tion between cohabitation and native born, tocontrol for the possible differences in behaviorassociated with nativity.

RESULTS

Descriptive Statistics

The distributions of the independent variables areshown in Table 1 for the entire sample and sepa-rately by the parents’ union status at the child’sbirth. For time-varying covariates, the baselinemeasure is shown. Table 1 shows that marriedmothers are more often White and less frequentlyBlack, whereas cohabiting mothers are more fre-quently Black rather than White. Married and co-habiting mothers are equally likely to be MexicanAmerican.

Married mothers are about 5 years older thancohabiting mothers and are more likely to comefrom a married family and to attend a religiousservice weekly. Consistent with prior research,married parents have significantly more educa-tion and income than cohabiting parents. Approx-imately 20% of married fathers have less thana high school diploma compared to almost twofifths of cohabiting fathers, and the difference in


college education is even greater. Over 35% ofmarried fathers in this sample have a collegedegree compared to fewer than 4% of cohabitingfathers. In terms of mothers’ education, mostmothers have similar levels of education as the

father, but this is especially true for married moth-ers (64% compared to 53%, respectively).Greater percentages of cohabiting mothers areless educated than their male partners. The dif-ferent education levels between married and

Table 1. Distribution of Independent Variables by Union Status of Parents at Child’s Birth

Total (N ¼ 2,249) Married (n ¼ 886) Cohabiting (n ¼ 1,363)

Mother’s race or ethnicity

White 44.03 55.39 18.41a

Black 33.85 22.99 58.35a

Mexican American 22.12 21.62 23.24

Mothers’ characteristics

Age (years) 28.12 29.58 24.83a

Parents married at age 15 54.42 62.32 36.59a

Attend religious service weekly 24.19 27.27 17.27a

Education

Fathers’ education

Less than high school 25.60 20.59 36.88a

High school 27.20 22.06 38.78a

Some college 20.31 20.10 20.77

College 26.89 37.23 3.57a

Mothers’ education

Same as father 60.42 63.67 53.07a

More than father 18.60 18.11 19.71

Less than father 20.98 18.22 27.22a

Economic resources

Household annual income 51,324 62,376 26,400a

Father’s employment in prior week 89.91 94.66 79.22a

Mother’s employment in prior year 69.51 70.48 67.33

Perceived social support

Able to borrow $200 from family 91.09 95.04 82.19a

Family will provide child care 92.98 94.83 88.80a

Relationship quality

Emotional support from father (1 – 3) 2.72 2.74 2.68

Disagreement (0 – 6) 1.74 1.64 1.94

Father hits or slaps 1.66 1.78 1.38a

Mothers’ substance abuse 3.08 2.56 4.26

Fathers’ substance abuse 2.32 2.10 2.82

Mothers’ relationship history

Prior marriage 6.94 5.82 9.47

Prior cohabitation 20.93 18.19 27.09a

Parents’ fertility history

Mother has prior child 22.14 13.15 42.42a

Father has prior child 19.81 12.74 35.77a

Couple has prior child 52.01 56.39 42.11a

Couple has new child within 3 years 35.23 36.35 32.71

Child conceived prior to union 6.53 4.16 11.85a

Relationship duration prior to birth (years) 3.86 5.78 2.43a

Note: Weighted using sampling weights by city to adjust for marital and nonmarital birth rates. Baseline values presented

for all variables except the couple has a new child within 3 years. Source: Fragile Families and Child Wellbeing Study.aCohabitors differ significantly from married at child’s birth at the p � .05 level.


cohabiting parents are likely reflected in thedifferent levels of household income. Althoughmarried mothers report over twice the annualincome as cohabitors, it is possible that cohabit-ing mothers are less aware of the incomes of theircohabiting partners and may be underreportingthe total household’s income. With regard toemployment, married fathers more often than co-habiting fathers were employed at the time oftheir child’s birth (95% vs. 79%), but there areno significant differences in married and cohabit-ing mothers’ employment rates in the year prior tothe child’s birth (70% and 67%, respectively).Married mothers report that they are more likelyto have access to money and child care from theirextended family members.

Married and cohabiting mothers report some-what similar levels of emotional support (2.74out of 3 compared with 2.68 for cohabitors;p ¼ .09), but cohabiting mothers report signifi-cantly higher levels of disagreement in the monthprior to their child’s birth. Very few mothersreport domestic violence or substance abuseproblems, but it is possible that these figures areunderreported given that the mothers were askedabout these issues shortly following childbirth.

With regard to relationship history, marriedmothers are equally likely as cohabiting mothersto have been married before but less likely to havelived with another cohabiting partner prior totheir current union. In terms of fertility, marriedparents more often than cohabiting parents shareanother biological child (56% vs. 42%, respect-ively), yet cohabiting mothers and fathers morefrequently than married parents have childrenfrom previous relationships. Similar proportionsof married and cohabiting parents have anotherchild within 3 years, and cohabiting parents are

almost three times more likely than married pa-rents to have conceived the focal child prior totheir current union. Finally, married parents havebeen together over 3 years longer than cohabitingparents prior to the birth of their child (5.78 vs.2.43 years, respectively).

Life-Table Estimates

The life-table estimates shown in Table 2 andFigure 1 present the cumulative proportion ofchildren who experience their parents’ separationby age 3. By the end of the child’s third year,almost half (49%) of children born to cohabitingparents have experienced their parents’ separa-tion compared with just over 11% of childrenborn to married parents. In fact, almost 11% ofchildren born to cohabiting parents have experi-enced their parents’ separation within their first6 months (see Figure 1). Our findings are similarto the findings of Manning et al. (2004), yet weshow somewhat higher rates of separation for co-habiting parents by age 1 (22% vs. 15% in theirstudy). Cohabiting unions are increasingly unsta-ble (Bumpass & Lu, 2000), which might explainwhy our estimates are higher. Also, some of thedifference may stem from the Fragile Familiesdata focusing only on parents in urban areas.

Figure 2 shows the cumulative proportion ofunions ending in separation, separately for Black,White, and Mexican American parents (race orethnicity is defined by the mother’s race or ethni-city). These results are also presented in Table 2.Consistent with results shown in Figure 1, chil-dren born to married parents, regardless of raceor ethnicity, experience greater parental stabilitythan do children born to cohabiting parents. Thedifference in marital and cohabiting stability,

Table 2. Cumulative Proportion of Unions Ending in Separation by Union Status at Child’s Birth and Race or Ethnicity

Union Status

at Child’s Birth

Proportion Separated by 36 Months After Child’s Birth

Total (N ¼ 2,249) White (n ¼ 764) Black (n ¼ 990)

Mexican

American (n ¼ 495)

Married at birth (n ¼ 886) 11.1 5.8 27.2a

9.3

Cohabiting at birth (n ¼ 1,363) 48.7 45.4 57.4b

28.8

Total (N ¼ 2,249) 23.1 10.9 43.9c

15.6

Note: Results based on life-table estimates. Source: Fragile Families and Child Wellbeing Study.aProportion separating is significantly higher than White or Mexican American married parents at the p � .05 level.

bPropor-

tion separating is significantly higher than Mexican American cohabiting parents but statistically similar to White cohabiting

parents at the p � .05 level. Separation rates for White cohabiting parents are statistically similar to Mexican American cohabi-

tors.cProportion separating is significantly higher than White or Mexican American parents at the p � .05 level.


however, is smaller for Black children than it isfor White or Mexican American children becauseof the high rates of separation of Black marriedparents (27%), as compared with White (6%)and Mexican American (9%) married parents.Black children born to cohabiting parents aremore likely to experience their parents’ separa-tion (57%) than their Mexican American counter-parts (29%), but they are (statistically) equally

likely as White children born to cohabiting pa-rents (45%) to experience their parents’ sep-aration. Stability in the unions of MexicanAmerican cohabiting parents does not differ sig-nificantly from White cohabiting parents.

Multivariate Analysis

Table 3 shows the association between union sta-tus at the child’s birth and the hazard or relativerisk of parental separation by age 3. The bivariateresults (data not shown) indicate that for childrenborn to cohabiting parents, the risk of experienc-ing their parents’ separation by age 3 is 5.57 timesthat of children born to married parents. Control-ling only for race or ethnic differences in unionstability (Model 1) reduces the relative risk to3.87, but it is still statistically significant. Consis-tent with the results shown in Table 2 and Fig-ure 2, Mexican American and White parentshave similar risks of separation (hazard ratio ¼1.06; p ¼ .860), and Black parents are signifi-cantly more likely than White parents to sepa-rate (hazard ratio ¼ 2.66; p ¼ .000). As statedpreviously, this difference in union stabilitybetween Black and White parents is drivenlargely by differences in marital instability(27% vs. 6%, respectively) rather than differen-ces in cohabiting stability (57% vs. 45%,respectively) at the observed level.

FIGURE 1. CUMULATIVE PROPORTION OF UNIONS

ENDING IN SEPARATION.

0

10

20

30

40

50

60

0 1 6 11 16 21 26 31 36 41

Months from Child's Birth

Perc

ent

Sepa

rate

d

Married Cohabiting

Note: Life-table estimates weighted using sampling weights

by city to adjust for marital and nonmarital birth rates. Married

and cohabiting refer to parents’ relationship status at child’s

birth. Source: Fragile Families and Child Wellbeing Study.

FIGURE 2. CUMULATIVE PROPORTION OF UNIONS ENDING IN SEPARATION BY RACE OR ETHNICITY.

0

10

20

30

40

50

60

70

0 1 6 11 16 21 26 31 36 41

Months from child's birth

Percen

t S

ep

arated

Married White Cohabiting White Married Black

Cohabiting Black Married Hispanic Cohabiting Hispanic

Note: Life-table estimates weighted using sampling weights by city to adjust for marital and nonmarital birth rates. Married

and cohabiting refer to parents’ relationship status at child’s birth. Source: Fragile Families and Child Wellbeing Study.


Table 3. Hazard Ratios of Parental Separation by Year 3 (N ¼ 3,699 Person-Years)

(1) (2) (3) (4) (5) (6)

Union status at child’s birth

(Married)

Cohabiting 3.87** 2.66** 2.43** 2.84** 2.57** 2.59**

Relationship duration prior to birth (years) 0.97 0.97 0.97 0.95 0.95

Mother’s race or ethnicity

(White)

Black 2.66** 2.30** 2.01** 2.01** 2.19** 1.72*

Mexican American 1.06 0.79 0.72 0.78 0.81 0.71

Mother’s characteristics

Age 0.97 0.98 0.97* 0.97 0.97

Parents married at age 15 0.75 0.76 0.80 0.81 0.86

Attend religious service weekly 0.98 0.96 1.09 1.06 1.11

Education

Father’s education

(Less than high school)

High school 0.84 0.95 0.89 0.83 0.95

Some college 0.66 0.88 0.82 0.67 0.99

College 0.57 0.92 0.74 0.53 0.92

Mother’s education

(Same as father)

More than father 0.89 0.97 0.95 0.89 0.97

Less than father 1.45 1.29 1.36 1.68y 1.43

Economic resources

Household annual income 0.99* 0.99*

Father’s employment 1.12 1.19

Mother’s employment 0.98 0.98

Perceived social support

Able to borrow $200 from family 1.21 1.28

Family will provide child care 0.78 0.89

Relationship quality

Emotional support (1 – 3) 0.68* 0.73yDisagreement (0 – 6) 1.13* 1.13*

Father hits or slaps 1.39 1.00

Mothers’ substance abuse 1.59 1.64

Fathers’ substance abuse 1.72y 1.89*

Relationship history

Prior marriage 1.66y 1.43

Prior cohabitation 0.74 0.75

Parents’ fertility history

Couple has prior child 0.82 0.79

Mother has prior child 0.93 0.85

Father has prior child 1.07 1.08

Couple has new child within 3 years 0.47** 0.47**

Child conceived prior to union 0.54* 0.55

�2 log-likelihood 3213.29 3185.6 3170.2 3153.7 3155.3 3111.4

Note: Results derived from Cox proportional hazard models. Time-varying covariates include annual household income, fa-

thers’ and mothers’ employment, perceived social support, emotional support, father hits or slaps, mothers’ and fathers’ sub-

stance abuse, and couple has new child. Reference category is in parentheses. Source: Fragile Families and Child Wellbeing

Study.

yp � .10. *p � .05. **p � .01.


Model 2 adds in controls for the duration of theparents’ relationship prior to the child’s birth,background characteristics of the mother, andthe parents’ education levels. These factorsjointly reduce the relative risk of separation forcohabiting versus married parents to 2.66 from3.87 in Model 1. Although none of the additionalcovariates in Model 2 is statistically significant,the inclusion of the duration and education vari-ables significantly increase the fit of the model,determined by likelihood ratio tests. In addition,these factors, particularly differences in educa-tion levels, reduce the relative risk of separationfor Blacks versus Whites (the hazard ratio corres-ponding to Black declines from 2.66 to 2.30).

Economic characteristics of the parents andperceived social support are included in Model3. These measures significantly add to the fit ofthe model (compared with Model 2) and attenuatethe relative risk for cohabitation to 2.43. Eco-nomic characteristics also account for a substan-tial amount of the difference in the hazard ofseparation between Black and White parents(the hazard ratio representing Black declinesfrom 2.30 to 2.01 with the inclusion of the pa-rents’ economic characteristics). Annual house-hold income predicts separation, with higherincomes related to a lower risk of separation. Asshown in Table 1, married parents have overtwice the annual household income as cohabitingparents have. We also find that fathers’ and moth-ers’ employment are not significantly related toseparation, after accounting for education andincome levels. In preliminary models, we con-trolled for mothers’ and fathers’ earnings insteadof annual income and employment. The resultsare robust to either method, and we chose to pres-ent the more parsimonious model. It is also likelythat income is representative of the length and sta-bility of employment in addition to the wagesearned. Our measures of perceived social supportare not significantly associated with separation.

Measures of the parents’ relationship qualityare included in Model 4. Consistent with priorresearch, emotional support is protective againstseparation (Carlson et al., 2004), and higher lev-els of disagreement and paternal substance abusehasten the parents’ breakup. Although thesemeasures significantly add to the fit of the model(compared with Model 2), this set of covariatesdoes not reduce the relative risk of cohabitationas we predicted; rather the relative risk for cohab-itation actually increases slightly between Mod-els 2 (2.66) and 4 (2.84). Supplemental analyses

in which we interacted the relationship qualityvariables with union status showed that the ef-fects of the relationship quality measures aresomewhat stronger for cohabiting versus marriedparents. Cohabiting parents appear to be moresensitive to the quality of their relationships andmore likely than married parents to end a relation-ship that involves conflict or substance abuse.Married parents appear to be more willing toremain together, despite problems in their rela-tionship, which is consistent with the idea thatmarriage is a stronger commitment than coha-bitation and that there are fewer barriers to exita cohabiting relationship versus marriage. Inter-estingly, relationship quality (particularly emo-tional support and disagreement) explains asimilar amount of the Black or White differencein separation as economic resources do.

Model 5 includes measures of family complex-ity, including mothers’ relationship history and theparents’ fertility histories. Including these varia-bles significantly adds to the fit of the model butdoes little to account for the difference in the rela-tive risk of separation for cohabiting and marriedparents (compared with Model 2). Parents whohave been married previously are more likely toseparate; married parents are more likely than co-habiting parents, however, to have experienceda prior marriage, which explains why this variabledoes not account for the difference in the risk ofseparation. In addition, having another child sub-sequent to the focal child’s birth is protectiveagainst separation, but cohabiting and married pa-rents are equally likely to have an additional child.Conceiving the focal child prior to the union statusat birth is also protective against separation, andcohabitors are significantly more likely than mar-ried parents to have formed their union followingconception. Interestingly, the family complexitymeasures do not explain as much of the Black orWhite difference in separation as economic re-sources and relationship quality measures explain.The coefficient for Black declines from 2.30 inModel 2 to 2.19 in Model 5, but it declined to2.01 with the inclusion of the economic and rela-tionship quality measures (Models 3 and 4).

Model 6 is the full model and includes all thesets of covariates mentioned above. Contrary towhat we predicted, combining the three sets ofcovariates (economic resources, relationshipquality, and family complexity) does not mediatethe difference in the relative risk of separation forchildren born to cohabiting and married parents.Indeed, we find that the hazard ratio is reduced


more by controlling only for differences in unionduration, background characteristics, education,and economic resources (Model 3), rather thanincluding controls for relationship quality andfamily complexity. Controlling for relationshipquality differences actually increases the risk ofseparation for cohabiting relative to marriedparents because cohabitors are more likely toseparate given a poor relationship, and they havesomewhat lower relationship quality. Althoughcohabitors have more complex families than mar-ried parents, these factors are not as strongly asso-ciated with union stability and thus do little toattenuate the difference in separation.

By contrast, including economic, relationshipquality, and family characteristics does help toaccount for the difference in the hazard of separa-tion for Black as compared with White parents(the coefficient declines from 2.30 in Model 2to 1.72 in Model 6, and the fit of the model is alsosignificantly improved). The race gap, however,is not completely mediated by the inclusion ofthese indicators in the model.

Although we have accounted for over half(54%) of the observed relative risk of separationbetween cohabiting and married parents (the haz-ard ratio declined from a bivariate of 5.57 to 2.59in Model 6), we have not accounted for all theobserved difference in separation. Parents whoare cohabiting at their child’s birth still have overtwo and a half times the risk of separating ascompared with parents who are married at theirchild’s birth, after controlling for an extensiveset of covariates. Model 6 also suggests that theeconomic, relationship quality, and family com-plexity variables have largely independent effectson separation, as indicated by the small change inthe size of the respective coefficients when allvariables are included in the same model as com-pared with the model in which only a particularset of covariates is included. Children who haveparents with more household income and bet-ter relationship quality and parents who haveanother child within 3 years of the focal child’sbirth have a significantly lower risk of experien-cing their parents’ separation. Children with pa-rents who argue often and who have substanceabuse problems experience a higher risk of sepa-ration, all else equal.

Racial and Ethnic Differences

Race-ethnic differences in the relative risk of sep-aration persist after the inclusion of an array of

demographic, economic, relationship quality,and family complexity variables. These variablesaccount for a significant portion of the racial gap,yet children born to Black mothers as comparedwith White mothers have a 72% higher risk ofexperiencing their parents’ separation by year 3,all else equal. As stated previously, the Black orWhite gap in union disruption is greater amongwomen who are married than cohabiting (seeFigure 2; Table 2). Investigative analysis showsthat with no additional controls, Black married pa-rents have 5.33 times the risk of separating within3 years following their child’s birth relative toWhite married parents, whereas Black cohabitorshave a similar risk of separation as White cohab-itors over this time frame. With the inclusion ofthe controls in Model 6, the racial gap in marriageinstability becomes more similar, yet marriedBlacks still have 2.95 times the risk of separatingas married White parents (results not shown).

To further investigate race and ethnic differen-ces, we ran each of our models separately by raceand ethnicity. As discussed above, this strategy issupported by our Chow tests. Our goal was todetermine whether the factors that explain the dif-ferences between marital and cohabiting stabilityare similar for each race and ethnic group. Wepresent two panels in Table 4: the first panelshows the relative risk of separation for cohabit-ing versus married parents separately for White,Black, and Mexican American parents regardlessof nativity status; the second panel shows theresults for Mexican American parents only andincludes controls for native-born status and aninteraction between native born and cohabitation.

Because of limitations in cell sizes, we did notcreate separate groups for native- and foreign-born Mexican Americans. In effect, the coeffi-cient associated with cohabitation in Panel Brepresents the risk of separation for foreign-borncohabitors relative to foreign-born married pa-rents; the coefficient associated with native bornillustrates whether there are differences in the riskof separation between native-born married pa-rents and foreign-born married parents; and thecoefficient for the interaction term illustrateswhether the risk of separation for native-bornand foreign-born cohabitors is similar.

At the bivariate level, all cohabiting parentshave a higher risk of separation than do marriedparents, regardless of race or ethnicity. The dif-ference in stability between cohabiting andmarried parents is particularly large for Whiteparents, however. White cohabiting parents have


almost 10 times the risk of separation by theirchild’s third year as compared with White mar-ried parents, whereas Mexican American cohab-iting parents have 3.56 times the risk and Blackcohabiting parents have 2.65 times the risk ofseparating relative to their married counterparts.The relative risk of separation is significantlydifferent for White versus Black and MexicanAmerican cohabitors, determined by results froma model interacting only race and cohabitation.Note that Black parents, however, have the high-est disruption rates overall, particularly for mar-ried parents, whereas Whites and MexicanAmericans have similar lower dissolution levels(see Table 2; Figure 2).

Panel A in Table 4 shows the difference in therisk of separation for cohabiting versus marriedparents across the series of models. Differencesin demographic characteristics explain a consid-erable portion of the difference in separation foreach race or ethnic group, but this is particularlytrue for Whites. Accounting for differences inrelationship duration prior to the child’s birth,maternal background characteristics, and pa-rents’ education levels reduces the hazard ratioon cohabitation for White parents from 9.79 to2.35 and significantly adds to the fit of the model,using a likelihood ratio test (data not shown). Bycontrast, these variables account for a smallerreduction in the hazard ratio for Black parents(from 2.56 to 2.11) and for Mexican Americanparents (3.56 to 3.33), although the fit of the

model is improved for each group. Differencesbetween cohabiting and married parents in rela-tionship duration, maternal age, and educationare much larger for White parents as comparedwith Black or Mexican American parents, whichhelps to explain why these factors explain somuch more of the cohabitation effect for Whites(data not shown). For example, the relationshipduration differential is over 4 years for White pa-rents (6.14 years for White married comparedwith 1.94 years for White cohabiting parents),but it is fewer than 3 years for Black or MexicanAmerican parents. Similarly, the age differentialis over 7 years for White parents, but it is fewerthan 4 years for Black parents and not statisticallysignificant for Mexican American parents. Interms of education, 31% of White cohabiting fa-thers have less than a high school degree, com-pared with only 4.5% of White married fathers;for Blacks, the corresponding numbers for fatherswith less than a high school degree are 29% ver-sus 14%, and for Mexican Americans, thereare no significant differences between married(69%) and cohabiting (61%) fathers; both groupshave very low levels of education.

Differences in economic resources (householdincome and employment) between cohabitingand married parents further reduces the differencein the risk of separation for each race and ethnicgroup, although the difference in log-likelihoodsis not significant for Whites, but this finding isparticularly true for Mexican Americans (see

Table 4. Hazard Ratio of Parental Separation for Cohabiting Versus Married at Birth Separately for Each Race or Ethnicity

Bivariate

Demographic

Characteristics

Economic

Resources

Relationship

Quality

Family

Complexity

Full

Model

(1) (2) (3) (4) (5) (6)

Panel A

White 9.79** 2.35* 2.33* 2.08* 2.07y 1.78

Black 2.65** 2.11** 1.94* 2.18** 2.11** 2.09*

Mexican American 3.56* 3.33* 2.71y 3.79** 3.47** 2.98**

Panel B

Mexican American Only

(Foreign born)

Cohabitation 1.09 1.25 0.96 1.16 1.14 0.78

Native born 0.60 0.51 0.55 0.34y 0.57 0.41

Interaction 7.29y 5.73y 5.94* 8.24** 7.64* 11.15**

Note: Cox proportional hazard models run separately by race. Each model in Panel A includes the same covariates, respec-

tively, as included in Table 3. In Panel B, the models add controls for native born and an interaction term between native born

and cohabitation status in order to isolate the hazard ratio of separation from cohabitation for foreign-born Mexican Americans.

Source: Fragile Families and Child Wellbeing Study.

yp � .10. *p � .05. **p � .01.


Model 3). Economic characteristics reduce thehazard ratio on cohabitation by 19% for MexicanAmericans but only 8% for Blacks and less than1% for Whites (Model 3 as compared with Model2). We find that income and paternal employmentare not the economic factors driving the unionstatus differential among Mexican Americans.One possible explanation is that maternal em-ployment seems to increase the risk of separationfor Mexican Americans, whereas it has no signif-icant effect on union stability for the othergroups. The difference in maternal employmentfor Mexican Americans by union status is negli-gible for foreign-born Latinas (and very few areemployed at their child’s birth or when the childis age one), but the union status differential islarge and significant for native-born MexicanAmericans when the child is age one (37% ofmarried mothers are employed as compared to66% of cohabiting mothers). Because native-born cohabitors are the most likely MexicanAmerican group to separate (as discussed below),this differential could explain why economic re-sources seem to explain more of the union statusdifferential for Mexican Americans than forWhites or Blacks.

By contrast to the economic characteristics,relationship quality characteristics explain a largeportion of the relative risk of separation for Whitecohabitors but explain none of the difference inseparation for Blacks and Mexican Americans.In fact, for Blacks and Mexican Americans, therelative risk of separation for cohabitors actuallyincreases relative to Model 2 with the inclusion ofthe relationship quality characteristics. The rea-son for this finding is likely twofold: first, differ-ences in relationship quality between married andcohabiting parents are substantially large forWhite parents but statistically insignificant forBlack and Mexican American parents; and sec-ond, the effects of relationship quality on separa-tion, particularly emotional support, are greaterfor White parents as compared with Black orMexican American parents. Again, we testedthese differences across race and ethnic groupsby running a fully interacted model (results notshown).

The family complexity measures yield some-what mixed results across race and ethnic groups.For White parents, these variables significantlyadd to the fit of the model and jointly reducethe hazard of separation for cohabitors from2.35 in Model 2 to 2.07 in Model 5. Higher levelsof prior marriage and children from prior unions

among White cohabiting versus married parentsaccount for this reduction in the hazard. Whitecohabitors are twice as likely as White marriedparents to have been married before and about10 times as likely to have children from priorunions (data not shown). There is no union gapon these variables for Black or Mexican Ameri-can parents, however. Thus, not surprisingly,the family complexity variables do not explainany of the cohabitation effect for Blacks orMexican Americans.

The inclusion of economic, relationship qual-ity, and family complexity covariates in Model6 explains almost a quarter of the relative riskof separation between White married and cohab-iting parents, relative to controlling for demo-graphic characteristics alone (Model 2), andover 80% of the higher risk of separation for co-habitors relative to the bivariate model (Model 1).Indeed, the combination of these variables ex-plains all the significant difference in instabilitybetween White married and cohabiting parents.Four variables are primarily responsible for thereduction in the relative risk between Models 2and 6: paternal substance abuse, previous mar-riages, and prior children from other unions.Thus, it is differences in relationship qualityand family complexity that mediate the differ-ence in stability between cohabiting and marriedWhite parents, controlling for the of backgroundcharacteristics of these families.

By contrast, the combination of these variablesdoes little to explain the higher risk of separationfor cohabiting Black and Mexican American pa-rents relative to their married counterparts. Eco-nomic factors make the largest contribution tothe fit of the model and seem to explain more ofthe difference in instability between marriedand cohabiting minority parents than the otherfactors.

The inclusion of the full set of covariates inModel 6 significantly reduces the differencesacross race or ethnic groups in the differentialseparation rates for married and cohabiting pa-rents relative to the bivariate models. Neverthe-less, despite the inclusion of these rich sets ofcovariates, the higher risk of separation for co-habiting parents relative to married parents per-sists for Black and Mexican American parents.This effect appears to be strongest for MexicanAmerican parents, yet the difference betweenBlack and Mexican Americans is not statisticallysignificant (on the basis of a fully interactedmodel).


Combining all Mexican American parents intoone group may be problematic, however, giventhat social and economic conditions often differby nativity status. Panel B in Table 4 presentsthe results for Mexican Americans only; eachmodel controls for nativity status and an interac-tion term between cohabiting at birth and beingnative born. The results suggest that the findingsfor Mexican Americans presented in Panel A ofTable 4 are largely reflective of native-born Mex-ican Americans but not of Mexican Americansborn in Mexico.

Foreign-born cohabitors are not more likely toseparate relative to foreign-born married parents(the reference group) at the bivariate level or aftercontrolling for any of the covariates included inthe models (none of the coefficients for cohabita-tion is significant at the p � .10 level). More-over, native-born married parents have similarrisks of separation as foreign-born married pa-rents, indicated by the lack of significance inthe hazard ratio for native born in each of themodels (with the exception of Model 4, whichis marginally significant). By contrast, the re-sults suggest that native-born cohabitors aremuch more likely than native-born marriedparents to separate within their child’s first3 years, and they are more likely to separatethan foreign-born cohabiting or married parents.Thus, native-born Mexican Americans are driv-ing the results for all Mexican Americansshown in Panel A of Table 4.

DISCUSSION

Recent increases in the percentage of childrenborn to cohabiting parents and growth in cohabi-tation make it important to provide timely assess-ments of how cohabitation relates to familystability. We extend prior work by focusing onthe relative stability of cohabiting and married pa-rents for a recent birth cohort of children and byexamining the factors that help explain the differ-ential rates of separation, paying close attentionto whether these processes differ by race andethnicity.

Our findings both mirror and move beyondprior studies; like those studies, we find that chil-dren born to cohabiting parents face a consider-ably higher risk of parental instability thanchildren born to married parents and that thisholds true across racial and ethnic groups. Cohab-itation appears to be particularly less stable forWhites, however. Three years following the birth

of a child, White cohabiting parents have almost10 times the risk of union disruption than Whitemarried parents. One explanation may be thatWhite mothers who have a birth while cohabitingare quite (negatively) selective (10% of Whitebirths are to cohabiting mothers, as compared toone fifth of Black and Hispanic [Bumpass &Lu, 2000]), whereas White mothers who havea birth while married are more common (76%of White births are marital, as compared to31% of Black births and 55% of Hispanic births[Hamilton et al., 2003]). Thus, there are greaterdifferences between White married and cohabit-ing mothers as compared with married and co-habiting Black or Mexican American mothersin economic and sociodemographic characteris-tics related to union instability.

Our second goal was to introduce potential fac-tors, specifically economic resources, social sup-port, relationship quality, and family complexitymeasures, that may help explain why childrenborn into cohabiting families face a higher riskof experiencing their parents’ breakup than thoseborn to married parents. Prior work has lackeddetailed measures of income and economic well-being, social support, relationship qualities, his-tories of relationship instability, and prior fertilitybehavior.

One of our key findings is that, on the whole,cohabiting parents are more likely to separateeven after accounting for differences in de-mographic, economic, relationship quality, andfamily complexity indicators. These indicatorsexplain over half of the observed difference inunion stability, but cohabiting parents remainover two and a half times as likely to separateas their married counterparts. We find that differ-ences in economic resources explain a greatershare of the cohabitation effect than the relation-ship quality or family complexity indicators. Allthree sets of indicators add to the fit of the model,however, and appear to be operating indepen-dently of one another.

Although we include economic characteristicsof the biological father, indicators of relationshipquality and family complexity, and a richer arrayof demographic characteristics in our models ascompared with prior studies, our results arelargely similar to those of Manning et al. (2004)who include only mothers’ demographic and eco-nomic characteristics. We find that children bornto cohabiting versus married parents have a 184%higher risk of experiencing their parents’ separa-tion by age three, in comparison to a 119% higher


risk by age 10 in the study of Manning et al. At thesame time, our multivariate analyses account fora much larger portion of the union status differen-tial relative to our bivariate models. The higherrates of cohabiting versus marital instability inthe current study may be explained by the fact thatcohabiting unions are increasingly less stable andthat our sample focuses primarily on lowerincome, urban residents who might have less sta-ble relationships than the general population.

One of the new findings in our study is thatamong Whites, the gap in the instability of cohab-iting and marital unions can be explained bydemographic, relationship quality, and familycomplexity indicators. The higher gap in unionstability for White parents is consistent with priorwork (Manning et al., 2004); the study byManning et al., however, did not explain theeffect of cohabitation among Whites. In this anal-ysis, parents’ education level, fathers’ substanceabuse, prior marriage, and children from priorunions explain the significant differences inunion stability between married and cohabitingWhite parents. White cohabiting parents havemuch lower levels of education and higher pater-nal substance abuse and are more likely to havebeen married previously and to have childrenfrom a prior union than White married parents.These factors are largely exogenous to their cur-rent relationship status, which implies that therisk of greater instability among White cohabitorsrelative to their married counterparts is largely theresult of selection into cohabitation versus mar-riage, rather than something caused by the rela-tionship status per se.

In contrast, controlling for relationship qualityand family complexity measures actually exacer-bates the difference in union stability for Blacksand Mexican Americans, and the socioeconomicand economic variables explain only a small por-tion of the cohabitation effect. Our analyses findthat among Black and Mexican American pa-rents, there are few differences between cohabit-ing and married parents in the measures includedin this analysis and that the effect of these meas-ures on instability is relatively weak. By contrast,there are considerable differences between Whitemarried and cohabiting parents on these indica-tors, and the effect of these indicators on instabil-ity is stronger. Prior work on married parentssuggests that the factors related to marital stabil-ity differ across race or ethnic groups (Sweeney &Phillips, 2004), and future work needs to explore

more fully how the factors differ for cohabitingparents as well.

Another contribution is the finding that amongforeign-born Mexican Americans (who representapproximately half of our sample of MexicanAmericans), we find no difference in the risk ofseparation between cohabiting and married pa-rents at the bivariate level. Among second orhigher generation Mexican Americans, however,the differences in union stability between cohab-iting and married parents is considerable andactually higher than the differential for Whitesand Blacks after accounting for an extensive setof covariates. To our knowledge, no study hasshown that the greater instability of cohabitingrelationships among Mexican Americans isdriven by the behaviors of native-born MexicanAmericans and that there are no differences in co-habiting and marital union stability among for-eign-born Mexican Americans. Prior researchthat has disaggregated foreign- and native-bornMexican Americans has focused on marital sta-bility and the risk factors (Phillips & Sweeney,2006), including premarital cohabitation (Phillips& Sweeney, 2005), that lead to greater disruption.That research finds significant differences bynativity status; marriages are more stable amongforeign-born Mexican Americans relative to theirnative-born counterparts (Phillips & Sweeney,2006), and premarital cohabitation seems to havea greater negative effect on marriage stability forforeign born as compared with native-born Mex-ican Americans (Phillips & Sweeney, 2005). Ourresearch shows that foreign-born cohabiting rela-tionships are also more stable than cohabiting re-lationships of native-born Mexican Americans,despite similar low levels of economic resources.

This finding highlights the importance of con-sidering family processes separately by nativitystatus. It is possible that cohabitation representsmore of an informal marriage to foreign-bornMexican American cohabitors on the basis oftheir cultural roots (Castro Martin, 2002) and thatcohabitation is a less committed union for native-born Mexican Americans. Oropesa and Landale(2004) find that foreign-born Hispanic womenhave more pronuptial attitudes and behaviorsthan native-born Hispanics and that subsequentgenerations adopt the social behaviors moreclosely associated with their socioeconomic sta-tus. Thus, the significance and nature of cohabita-tion may differ by nativity status (Brown et al.,2006) and needs further consideration.


There are several limitations to these analyses.One limitation is that we are unable to examinewhether marriage among cohabiting parents en-hances union stability. Prior work suggests thatmarriage promotes stability, but the protectiveeffect of marriage differs for race and ethnicgroups (Manning et al., 2004). Our data do notidentify mothers who transition to marriage andthen separate between the Wave 2 and Wave 3 in-terviews, precluding our ability to track all transi-tions into and out of marriage among cohabitors.Another limitation is that we attempt to incorpo-rate a broader range of observed characteristicsthan has been used in prior work, but some ofthe measures available in the Fragile FamiliesStudy may not be ideal indicators of what wedesire to measure. For example, perceived socialsupport is limited to financial aid and child care,employment does not indicate length or stabilityof employment, and relationship quality focusesprimarily on negative aspects of the relationship.The mechanisms underlying the union status gapin family stability may be better understood withimproved measurement and insights gleanedthrough qualitative methods. Refined measuresmight be particularly necessary for Black andMexican American parents as the measuresincluded in this analysis do not predict union sta-bility similarly for all groups. Another limitationis that we are unable to determine a causal linkbetween union status and stability, and it is prob-able that unobserved or unmeasured characteris-tics are omitted from this analysis.

Our findings contribute to the ongoing effort tobetter understand the role that cohabitation playsin family formation, and how the nature of cohab-itation may differ across race and ethnic groups.The growing consensus is that cohabitation isa temporary relationship for White couples thatis generally short lived and often precedes mar-riage; by contrast, cohabitation is viewed moreas an alternative to marriage for Black and manyMexican American parents who are unableto make a successful transition to marriage(Manning & Landale, 1996; Phillips & Sweeney,2005). Our findings support this in that at thezero-order level Black and Mexican Americancohabitors have more similar union stability astheir married counterparts than Whites. Yet, ourfindings indicate that understanding of cohabita-tion and family stability needs to be morenuanced to take account of nativity status (Brownet al., 2006). Important to note is that although thedifference in union stability between cohabiting

and married parents is greatest for Whites, Whitesstill have very low rates of instability, particularlymarital instability. Mexican Americans havea moderate differential at the bivariate level be-cause of low cohabitation instability, driven bythe low cohabitation instability of foreign-bornMexican Americans. In contrast, Blacks evi-dence a smaller cohabitation-marriage differen-tial because of high marital instability. Thus, themeaning and significance of marriage may differby race or ethnicity and nativity status, and it isessential to understand these differences morefully in order to adequately grasp the meaningof cohabitation relative to marriage for variousracial and ethnic groups.

Our findings also contribute to knowledgeabout the ramifications of cohabitation for chil-dren. The marital status of two biological parentsat the time of a child’s birth has implicationsfor the stability of children’s early family lifecourse. Researchers examining the implicationsof cohabitation should account for the greaterinstability experienced by children born to cohab-iting parents and should further explore to whatextent a transition to marriage from cohabitationenhances relationship stability.

This article speaks to current policy initiativesaimed at promoting marriage among unmarriedparents by providing relationship counseling.Our findings show that although relationshipquality is an important predictor of union stabil-ity, it does little to help explain the difference ininstability between cohabiting and married pa-rents for Black and Mexican American parents.Thus, policies aimed at promoting stabilityamong parents must take race or ethnic differen-ces into account; for minorities, economic resour-ces are better predictors of union stability.

Although our knowledge is expanding, the im-plications of cohabitation for children are stilllargely undefined, despite the increasing likeli-hood that a child will be born into a cohabitingunion. An important difference between thesechildren and those born to married parents is thatthey are significantly more likely to experiencetheir parents’ separation early in their life course,which may have serious consequences for theirsubsequent development. In addition, these fam-ily experiences and processes differ considerablyacross race and ethnic groups and by nativity sta-tus. Thus, any deleterious consequences associ-ated with the increase in cohabiting births mayexacerbate the existing inequalities in children’slife trajectories.


REFERENCES

Allison, P. (1984). Event history analysis. Beverly

Hills, CA: Sage.

Amato, P. (2005). The impact of family formation

change on the cognitive, social, and emotional

well-being of the next generation. Future of Chil-dren, 15, 75 – 96.

Antonovics, K., & Town, R. (2004). Are all the good

men married? Uncovering sources of the marital

wage premium. American Economic Review, 94,

317 – 321.

Becker, G. (1991). A treatise on the family. Cam-

bridge, MA: Harvard University Press.

Brines, J., & Joyner, K. (1999). The ties that bind:

Principles of cohesion in cohabitation and marriage.

American Sociological Review, 64, 333 – 355.

Brown, S. L. (2000). Union transitions among cohab-

itors: The role of relationship assessments and ex-

pectations. Journal of Marriage and the Family,

62, 833 – 846.

Brown, S. L. (2003). Relationship quality dynamics

of cohabiting unions. Journal of Family Issues, 24,

583 – 601.

Brown, S. L. (2004a). Family structure and child

well-being: The significance of parental cohabitation.

Journal of Marriage and Family, 66, 351 – 367.

Brown, S. L. (2004b). Moving from cohabitation to

marriage: Effects on relationship quality. SocialScience Research, 33, 1 – 19.

Brown, S. L., & Booth, A. (1996). Cohabitation ver-

sus marriage: A comparison of relationship quality.

Journal of Marriage and the Family, 58, 668 – 678.

Brown, S. L., Van Hook, J., & Glick, J. (2006). Gen-erational differences in cohabitation and marriagein the US (Working Paper Series 06 – 03). Bowl-

ing Green, OH: Center for Family and Demo-

graphic Research, Bowling Green State University.

Bumpass, L., & Lu, H. (2000). Trends in cohabitation

and implications for children’s family contexts.

Population Studies, 54, 29 – 41.

Carlson, M., & Furstenberg, F. (2006). The preva-

lence and correlates of multipartnered fertility

among urban US parents. Journal of Marriage andFamily, 68, 718 – 732.

Carlson, M., McLanahan, S., & England, P. (2004).

Union formation in fragile families. Demography,41, 237 – 262.

Casper, L., & Bianchi, S. (2002). Continuity andchange in the American family. Thousand Oaks,

CA: Sage.

Castro Martin, T. (2002). Consensual unions in Latin

America: Persistence of a dual nuptiality system.

Journal of Comparative Family Studies, 33, 35 – 55.

Cherlin, A. (2004). The deinstitutionalization of Amer-

ican marriage. Journal of Marriage and Family, 66,

848 – 861.

Cohen, P. N. (2002). Cohabitation and the declining

wage premium for men. Work and Occupations,29, 346 – 363.

DeMaris, A. (2000). Till discord do us part: The role

of physical and verbal conflict in union disruption.

Journal of Marriage and the Family, 62, 683 – 692.

DeMaris, A. (2002). Regression models. In M.

Wiederman & B. Whitley (Eds.), Handbook forconducting research on human sexuality (pp.

255 – 288). Mahway, NJ: Erlbum.

Eggebeen, D. (2005). Cohabitation and exchanges of

support. Social Forces, 83, 1097 – 1110.

Fein, D., Burstein, N., Fein, G., & Lindberg, L.

(2003). The determinants of marriage and co-habitation among disadvantaged Americans: Re-search findings and needs. Cambridge, MA: Abt

Associates.

Fields, J. (2004). American families and living ar-rangements: 2003 (Current Population Reports,

Series P-20, No. 553). Washington, DC: U.S. Cen-

sus Bureau.

Gibson-Davis, C., Edin, K., & McLanahan, S.

(2005). High hopes but even higher expectations:

The retreat from marriage among low-income

couples. Journal of Marriage and Family, 67,

1301 – 1312.

Ginther, D. K., & Zavodny, M. (2001). Is the male

marriage premium due to selection? The effect of

shotgun weddings on the return to marriage. Jour-nal of Population Economics, 14, 313 – 328.

Goldscheider, F., & Sassler, S. (2006). Children and

stepfamily formation. Journal of Marriage andFamily, 68, 275 – 291.

Hamilton, B., Martin, J., & Sutton, P. (2003). Births:

Preliminary data for 2002. National Vital StatisticsReports, Center for Disease Control, 51(11).

Hao, L. (1996). Family structure, private transfers,

and the economic well-being of families with chil-

dren. Social Forces, 75, 269 – 292.

Harknett, K., & Knab, J. (2007). More kin, less sup-

port. Multipartnered fertility and perceived support

among mothers. Journal of Marriage and Family,

69, 237 – 253.

Kenney, C. T. (2003). Household economies: Money

management and resource allocation among mar-

ried and cohabiting parents. Dissertation AbstractsInternational, A: The Humanities and Social Sci-ences, 63(7), 2709-A.

Korenman, S., & Neumark, D. (1997). Does marriage

really make men more productive? Journal ofHuman Resources, 26, 282 – 307.


Landale, N., & Hauan, S. (1992). The family life

course of Puerto Rican children. Journal of Mar-riage and the Family, 54, 912 – 924.

Lundberg, S., & Pollack, R. A. (1994). Noncoopera-

tive bargaining models of marriage. AmericanEconomic Review, 84, 132 – 137.

Manning, W. D. (1993). Marriage and cohabitation

following premarital conception. Journal of Mar-riage and the Family, 55, 839 – 850.

Manning, W. D. (2001). Childbearing in cohabiting

unions: Racial and ethnic differences. Family Plan-ning Perspectives, 33, 217 – 223.

Manning, W. D. (2004). Children and the stability of

cohabiting couples. Journal of Marriage and Fam-ily, 66, 674 – 689.

Manning, W. D., & Brown, S. L. (2006). Children’s

economic well-being in married and cohabiting

parent families. Journal of Marriage and Family,

68, 345 – 362.

Manning, W. D., & Lamb, K. (2003). Parental cohab-

itation and adolescent well-being. Journal ofMarriage and Family, 65, 876 – 893.

Manning, W. D., & Landale, N. S. (1996). Racial and

ethnic differences in the role of cohabitation in pre-

marital childbearing. Journal of Marriage and theFamily, 58, 1 – 15.

Manning, W. D., & Smock, P. J. (2002). First comes

cohabitation, then comes marriage? Journal ofFamily Issues, 23, 1065 – 1087.

Manning, W. D., Smock, P. J., & Majumdar, D.

(2004). The relative stability of marital and cohab-

iting unions for children. Population Research andPolicy Review, 23, 135 – 159.

Marcil-Gratton, N., Le Bourdais, C., & Lapierre-

Adamcyk, E. (2000). The implications of parents’

conjugal histories for children. Canadian Journalof Policy Research, 1, 32 – 40.

Marks, N., & McLanahan, S. (1993). Gender, family

structure, and social support among parents. Jour-nal of Marriage and the Family, 44, 481 – 493.

McLanahan, S., & Casper, L. (1995). Growing diver-

sity and inequality in the American family. In

R. Farley (Ed.), State of the union: America in the1990s (Vol. 2, pp. 1 – 45). New York: Russell Sage.

McLanahan, S., & Sandefur, G. (1994). Growing upwith a single parent: What hurts, what helps?Cambridge, MA: Harvard University Press.

Musick, K. (2002). Planned and unplanned childbear-

ing among unmarried women. Journal of Marriageand Family, 64, 915 – 930.

Nock, S. (1995). A comparison of marriages and co-

habiting relationships. Journal of Family Issues,16, 53 – 76.

Oropesa, R. S., & Landale, N. S. (2004). The future

of marriage and Hispanics. Journal of Marriageand Family, 66, 901 – 920.

Osborne, C. (2005). Marriage following the birth of

a child among cohabiting and visiting parents.

Journal of Marriage and Family, 67, 14 – 26.

Osborne, C., & McLanahan, S. (2007). Partnership

instability and child well-being. Journal of Mar-riage and Family, 69, 1065 – 1083.

Osborne, C., McLanahan, S., & Brooks-Gunn, J.

(2005). Young children’s behavioral problems inmarried and cohabiting families (Working Paper

No. 2003-09-FF). Princeton, NJ: Center for Re-

search on Child Well-Being, Princeton University.

Pagnini, D., & Morgan, S. P. (1996). Racial differen-

ces in marriage and childbearing: Oral history

evidence from the South in the early twentieth

century. American Journal of Sociology, 101,

1694 – 1718.

Phillips, J. A., & Sweeney, M. M. (2005). Premarital

cohabitation and marital disruption among White,

Black, and Mexican American women. Journal ofMarriage and Family, 67, 295 – 313.

Phillips, J. A., & Sweeney, M. M. (2006). Can differ-

ential exposure to risk factors explain recent racial

and ethnic variation in marital disruption? SocialScience Research, 35, 409 – 434.

Raley, R. K., & Bumpass, L. L. (2003). The topogra-

phy of the plateau in divorce levels and trends in

union stability after 1980. Demographic Research,

8, 246 – 258.

Raley, R. K., Durden, T. E., & Wildsmith, E. M.

(2004). Understanding Mexican American mar-

riage patterns using a life course approach. SocialScience Quarterly, 85, 872 – 890.

Raley, R. K., & Wildsmith, E. M. (2004). Cohabita-

tion and children’s family instability. Journal ofMarriage and Family, 66, 210 – 219.

Roebuck Bulanda, J., & Brown, S. (2007). Race-

ethnic differences in marital quality and divorce.

Social Service Review. 36, 945 – 967.

Smock, P. J. (2000). Cohabitation in the United

States: An appraisal of research themes, findings,

and implications. Annual Review of Sociology, 26,

1 – 20.

Smock, P. J., & Manning, W. D. (1997). Cohabiting

partners’ economic circumstances and marriage.

Demography, 34, 331 – 341.

Smock, P. J., Manning, W. D., & Porter, M. (2005).

Everything’s there except the money: How eco-

nomic factors shape the decision to marry among

cohabiting couples. Journal of Marriage andFamily, 67, 680 – 696.


Stewart, S. D., Manning, W. D., & Smock, P. J.

(2003). Union formation among men in the US:

Does having prior children matter? Journal ofMarriage and Family, 65, 90 – 104.

Stratton, L. S. (2002). Examining the wage differen-

tial for married and cohabiting men. EconomicInquiry, 40, 199 – 212.

Sweeney, M. M., & Phillips, J. A. (2004). Under-

standing racial differences in marital disruption:

Recent trends and explanations. Journal of Mar-riage and Family, 66, 639 – 650.

Teachman, J. (2004). The childhood living arrange-

ments of children and the characteristics of their

marriages. Journal of Family Issues, 25, 86 – 111.

Waite, L. (2000). Trends in men’s and women’s

well-being in marriage. In L. Waite (Ed.), The ties

that bind: Perspectives on marriage and cohabita-tion (pp. 368 – 392). New York: Aldine De

Gruyter.

White, L., & Booth, A. (1985). The quality and

stability of remarriages: The role of step-

children. American Sociological Review, 50,

689 – 698.

Wildsmith, E., & Raley, K. (2006). Race-ethnic dif-

ferences in nonmarital fertility: A focus on Mexi-

can American women. Journal of Marriage andFamily, 68, 491 – 508.

Wu, L., Bumpass, L., & Musick, K. (2001). Histori-

cal and life course trajectories of nonmarital child-

bearing. In L. Wu & B. Wolfe (Eds.), Out ofwedlock: Causes and consequences of nonmaritalfertility (pp.3 – 48). New York: Russell Sage.


Married and Cohabiting Parents’ Relationship Stability: A Focus on Race and Ethnicity

Documents

Transcript of Married and Cohabiting Parents’ Relationship Stability: A Focus on Race and Ethnicity