Interracial Relationships and the Transition to Adulthood

Scholars have long viewed interracial mar-riage as an indicator of the social and geo-

graphic distance between different racial groups(Blau 1994 Bogardus 1967 Gordon 1964Lieberson and Waters 1988 Spickard 1989)1

Suggesting that barriers between racial groupshave eroded in recent decades data from the

US Census shows that the proportion of mar-ried couples with partners from different racialgroups has increased remarkably since the1960s Although the number of marriagesincreased only marginally between 1960 and2002 the number of interracial marriagesincreased from 157000 to 1674000 However

IInntteerrrraacciiaall RReellaattiioonnsshhiippss aanndd tthhee TTrraannssiittiioonn ttoo AAdduulltthhoooodd

Kara Joyner Grace KaoCornell University University of Pennsylvania

This study tracks and explains changing patterns of involvement in interracial sexual

relationships during the transition to adulthood Using a life course perspective that

highlights the role of historical changes as well as age-graded changes in contexts and

relationships the authors hypothesize that involvement in interracial sexual

relationships declines with increasing age among young adults The analyses are based

on some of the first nationally representative surveys to collect detailed information on

sexual relationships the National Longitudinal Study of Adolescent Health and the

National Health and Social Life Survey Findings from these surveys show that

individuals are decreasingly likely to be in an interracial relationship between the ages

of 18 and 35 years They also suggest that the age decline in interracial involvement is a

by-product of the transition to marriage in young adulthood and the increasing

formation of interracial relationships in recent years These findings have implications

for future research on interracial relationships and family formation

AAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW 22000055 VVOOLL 7700 ((AAuugguusstt556633ndashndash558811))

2462-ASR 704 filename70402-Joyner

Direct cor respondence to Kara JoynerDepartment of Policy Analysis and ManagementCornell University 109 MVR Hall Ithaca NY14853 (kj34cornelledu) Email for Grace Kaograce2sscupennedu This research was sup-por ted by new faculty grants from McGillUniversity Cornell University and the UniversityResearch Foundation of the University ofPennsylvania as well as a grant from the NationalInstitute of Child Health and Human Development(NICHD) (R01 HD38704-01A1) The authors arethankful for the helpful comments provided by theeditors and reviewers at ASR as well as those ofRachel Dunifon and Raymond Swisher The authorsalso thank the par ticipants at the CarolinaPopulation Center Weekly Seminar for their usefulcomments on preliminary analyses and partici-pants at the following Cornell seminars theSociology Colloquium the Human Development

Brown Bag Series and the Demography andInequality Forum The authors are indebted to Won-Hong Ung for his help in creating the contextualdatabase for National Health and Social Life Survey(NHSLS) respondents and to Pat Lauper for edi-torial assistance

1 Many scholars use the term ldquoracerdquo in referenceto groups socially defined on the basis of perceivedphysical differences (Cornell and Hartmann 1998)Social science studies including those on interracialmarriage typically use racial classifications of non-Hispanic white non-Hispanic black (or AfricanAmerican) Hispanic (or Latino) and non-HispanicAsian (Farley 1996 Harris and Ono 2004 Jacobs andLabov 2002 Qian 1997) Although Hispanic is apanethnic category composed of both racial and eth-nic groups studies typically include it as a separaterace category (Harris and Sim 2002 HirschmanAlba and Farley 2000)

Delivered by Ingenta toMr Matharu (cid 32629) ingenta internal live 231F4 (cid 75000325) University of Pennsylvania

Library (cid 80001333)Wed 26 Oct 2005 211018

interracial marriage still is relatively uncommonIn 2002 only 29 percent of all marriages wereinterracial (US Bureau of the Census 19982003)2

Since the 1960s men and women have beenpostponing the age at which they first marry andthey are increasingly likely to reach the age of35 years without marrying Furthermore theyare more likely than ever before to form a cohab-iting relationship or accumulate five or more sexpartners during the course of young adulthood(Bumpass and Lu 2000 Casper and Cohen2000 Laumann et al 1994) Although nation-ally representative studies of mate selection andfamily formation have broadened their scope toinclude cohabiting relationships (Raley 2001Smock 2000) they generally have disregardedsingle (eg dating) relationships (for excep-tions see Forste and Tanfer 1996 Waite andJoyner 2001)

The couple of representative studies thatexamine the racial mix of partners in both sin-gle and cohabiting relationships show that indi-viduals in these types of relationships are morelikely than married individuals to have a part-ner of a different race (Blackwell and Lichter2004 Laumann et al 1994) Although innova-tive these studies focus only on current or rel-atively recent relationships in the early or mid1990s Consequently we do not know the extentto which involvement in different types of inter-racial sexual relationships has increased in thepast decade Nor do we know why marriages areless likely than other types of sexual relation-ships to be interracial Researchers speculatethat individuals are increasingly selective withrespect to race (and other traits) as they transi-tion from single relationships to cohabitation tomarriage but they have not been able to exam-ine matching processes fully because of incom-plete data on the histories of sexual relationships(Blackwell and Lichter 2004)

Previous studies have yet to fully theorize orexamine how involvement in interracial rela-tionships changes with age In a study that doesaddress age differences in interracial marriageHeaton and Albrecht (1996) found that in both

1980 and 1990 the likelihood of an individualhaving a spouse of a different race declined sig-nificantly with age3 They argued that the agedecline mainly reflects the fact that youngerindividuals began their marriages in more recentyears when interracial marriage was more com-mon We predict that interracial involvementacross a broader range of sexual relationshipswill show an even greater decline with ageAfter all individuals are increasingly likely tobe married as they get older and marriages areless likely than single and cohabiting relation-ships to be interracial

If interracial marriage is considered to be socritical why is it worth investigating single andcohabiting relationships Like interracial mar-riage these other types of interracial relation-ships are indicators of the social barriersbetween racial groups but they also may con-tribute to the erosion of these barriers by allow-ing individuals to learn about variation withinracial groups and to question racial stereotypes(Kalmijn 1998) Additionally it is critical toconsider the racial mix of partners in single andcohabiting relationships because fertility rateswithin these relationships are increasing (Raley2001) Finally understanding other types ofinterracial relationships helps to enhance ourunderstanding of interracial marriages Kingand Bratter (2004) found that women are morelikely to select husbands across racial lines ifthey select first sex partners across these lines

To foreshadow the age decline we use a lifecourse perspective that highlights how rela-tionships are socially structured by age and his-torical context (Crosnoe 2000 Elder 1998Shanahan 2000) Our analyses are based onsome of the first nationally representative sur-veys to collect detailed information on sexualrelationships the National Longitudinal Studyof Adolescent Health (Add Health) and theNational Health and Social Life Survey(NHSLS)

Focusing on current sexual relationships thefirst part of our study uses data from both sur-veys to examine age differences in interracialinvolvement among young adults in two dif-ferent periods (roughly 1990 and 2000) Turning

556644mdashmdashndashndashAAMMEERRIICCAANN SSOOCCIIOOLLOOGGIICCAALL RREEVVIIEEWW


2 These census statistics are based on the follow-ing categories of race white black American IndianAsianPacific Islander and other Hispanics are notincluded as a separate category

3 The study by Heaton and Albrecht (1996) exclud-ed marriages that included a Hispanic or Asian part-ner



to sexual relationships formed by young adultsbetween the two periods the second part of ourstudy uses data from Add Health to addresshow the formation of interracial relationshipsdiffers by age and historical period Because weare able to examine the transition to marriageamong single and cohabiting partners we canexplain why marriages are less likely than othertypes of sexual relationships to be interracial

IINNTTEERRRRAACCIIAALL RREELLAATTIIOONNSSHHIIPPSS IINN AALLIIFFEE CCOOUURRSSEE PPEERRSSPPEECCTTIIVVEE

We predict that involvement in interracial rela-tionships will decline with increasing age on thebasis of a life course perspective that drawsattention to the role of historical changes (peri-od and cohort effects) as well as age-gradedtransitions to different contexts and relation-ships (age effects) Cohort effects refer to influ-ences associated with membership in a birthcohort whereas period effects refer to influ-ences across cohorts within a given historicalperiod and age effects refer to influences relat-ed to aging or movement through the life course(Glenn 1976) Whereas research on interracialmarriage typically highlights period or cohorteffects (Kalmijn 1993 Qian 1997) studiesinvestigating the educational mix of partnersin marriage have additionally emphasized ageeffects (Mare 1991) In the following sectionswe discuss how preferences and opportunitiesfor interracial relationships differ according toage period and cohort and the role that thesedifferences may play in the age decline in inter-racial involvement

PPRREEFFEERREENNCCEESS FFOORR

IINNTTEERRRRAACCIIAALL RREELLAATTIIOONNSSHHIIPPSS

Researchers have long argued and found thatsocial distance attitudes or preferences for con-tact with other racial groups are more restric-tive as the relationship is increasingly intimate(Bogardus 1967 Park 1924 Schuman et al1997) They have continued to distinguish mar-riages from other types of relationships becauseof the greater commitment required in mar-riage Marriage involves not only a publicacknowledgement of the relationship but alsoan intention to remain with the partner for a life-time To varying degrees it also assumes thatpartners will share family and friends pool

resources and have children (Blackwell andLichter 2004 Laumann et al 1994) Suggestingthat preferences for interracial marriage areweaker than preferences for other types of inter-racial sexual relationships studies find thatmarriages are less likely than single and cohab-iting relationships to be interracial (Blackwelland Lichter 2000 2004 Jepson and Jepson2002 Laumann et al 1994)

AGE EFFECTS In samples of ongoing rela-tionships older individuals are more likely thanyounger individuals to be in marriages andthey are less likely to be in single and cohabit-ing relationships In 2000 19 percent of indi-viduals between the ages of 15 and 19 yearswere married as compared with 19 percent ofthose between the ages of 20 and 24 years 472percent of those ages 25 to 29 years and 625percent of those ages 30 to 34 years (USBureau of the Census 2001a) Speculating thatthe greater racial similarity among married cou-ples reflects a winnowing process researcherspredict that interracial relationships are lesslikely than same-race relationships to eventuatein marriage (Blackwell and Lichter 2000 2004)Involvement in interracial relationships maydecrease in the transition to adulthood becausethe majority of individuals marry in this phaseof the life course4

Also in comparison with younger individu-als older individuals have formed their currentrelationships at older ages As individuals getolder they may be decreasingly likely to forman interracial relationship because of greateranticipation that their relationships will even-tuate in marriage We term this process ldquoantic-ipatory mate selectionrdquo which is analogous tothe broader concept of anticipatory socializationa process whereby individuals ease into roletransitions by taking on the values of the groupsto which they aspire (Merton 1964) In the case

IINNTTEERRRRAACCIIAALL RREELLAATTIIOONNSSHHIIPPSS AANNDD TTRRAANNSSIITTIIOONN TTOO AADDUULLTTHHOOOODDmdashmdashndashndash556655


4 If interracial relationships are less likely thansame-race relationships to become marriages theyalso may be more likely to dissolve Nationally rep-resentative studies suggest that interracial relation-ships may be less durable than same-racerelationships because of external pressures (Heaton2002 Heer 1974 Monahan 1970) However someof these studies are dated and their evidence is incon-clusive (Aldridge 1978)



of anticipatory mate selection individuals arethought to select partners they expect to bemore acceptable marriage partners and conse-quently favor same-race partners over differ-ent-race partners This tendency is expected toincrease during the course of young adulthood

PERIOD EFFECTS The relationships of olderindividuals tend to be longer in duration thanthose of younger individuals The decline ininterracial involvement with age may reflect aperiod effect because younger individuals haveformed their relationships more recently andduring a period of greater racial toleranceApproval of interracial marriage in the gener-al population has increased dramatically over thepast four decades In the late 1950s only about1 of every 20 whites approved of marriagesbetween whites and nonwhites By the late1990s more than 2 of 3 whites approved ofthese marriages (Schuman et al 1997) Even inthe late 1990s social distance attitudes liberal-ized considerably For instance the percentageof whites who favored a relative marrying ablack person increased from 17 to 23 percentbetween 1996 and 2000 (Krysan 2002)

COHORT EFFECTS Individuals of any givenage vary greatly with respect to the age andyear that they formed their current relation-ships At any specific age or within any specif-ic period individuals from later cohorts may bemore likely than individuals from earlier cohortsto form interracial relationships because of theirgreater preferences for interracial contactPrevious research has found that within histor-ical periods individuals from earlier cohortsdisplay less tolerant social distance attitudesthan their counterparts from later cohorts main-ly because they developed their attitudes in peri-ods that were less accepting racially Althoughdifferences are somewhat slight cohorts canbe ordered monotonically in terms of their socialdistance attitudes (Schuman et al 1997)

OOPPPPOORRTTUUNNIITTIIEESS FFOORR


According to structural perspectives individu-als prefer to associate with members of theirown racial group However the size of theirracial group relative to other racial groups influ-

ences their ability to satisfy this preferenceIndividuals from smaller racial groups have lessopportunity than individuals from larger racialgroups to form relationships with members ofthe same race (Blau 1994) Supporting this per-spective studies that directly examine how therelative sizes of different racial groups in statesand metropolitan areas influence rates of inter-racial marriage find that individuals have agreater chance of marrying interracially whentheir own racial group is smaller than otherracial groups (Blau Blum and Schwartz 1982Harris and Ono 2004 Rytina et al 1988) Giventhe effect of relative racial group size it is nocoincidence that differences in interracial mar-riage by racial group correspond to differencesin the relative sizes of racial groups Whitesare the largest racial group in the US popula-tion followed by blacks Hispanics and AsiansFocusing on married couples 20 to 29 years ofage in 1990 Qian (1997) found that 2 percentof whites 54 percent of blacks 369 percent ofHispanics and 639 percent of Asians were inmarriages with a partner of a different race

COHORT EFFECTS Individuals from latercohorts may be more likely than their counter-parts from earlier cohorts to form interracialrelationships as a consequence of their greaterracial diversity But although the racial diversi-ty of youth has increased over time high schoolscontinue to be racially segregated (Orfield et al1997 Reardon Yun and Eitle 2000) Previousstudies of interracial friendship in adolescencebased on Add Health suggest that economicand geographic segregation limits opportunitiesfor interracial contact considerably across andwithin educational institutions (Moody 2001Mouw and Entwisle forthcoming Quillian andCampbell 2003) Given the magnitude ofschool-based segregation cohort differences inopportunities for interracial contact may playonly a marginal role in the age decline in inter-racial involvement

AGE EFFECTS Individuals may be less likelyto form interracial relationships with increasingage because of reduced opportunities for inter-racial contact which reflect the declining edu-cational enrollment of blacks and Hispanicsbetween the ages of 14 and 17 years and the factthat blacks and Hispanics are much less likely





than whites and Asians to attend college (USBureau of the Census 2001b) Yet to the extentthat educational institutions are segregated byrace opportunities for interracial contact maychange little as individuals advance from onegrade to the next or as they make the transitionfrom high school to college (Jacobs 1997) Onceindividuals enter the workplace their opportu-nities for interracial contact may even increaseSuggesting this social networks formed in theworkplace tend to be more racially diverse thannetworks formed elsewhere (Marsden 1990McPherson Smith-Lovin and Cook 2001Reskin McBrier and Kmec 1999) Theincreased opportunities for interracial contactthat accompany the transition to work thenmay partially suppress the age decline in inter-racial involvement

EEXXPPEECCTTAATTIIOONNSS AANNDD OOVVEERRVVIIEEWW OOFF SSTTUUDDYY

In summary the age decline in interracialinvolvement is thought to reflect the selectionof relationships at different ages Individuals areincreasingly likely to be in a marriage rather thana cohabiting or single relationship as they getolder and marriages are less likely than singleand cohabiting relationships to be interracialAlso with increasing age individuals have rela-tionships that are longer in duration These rela-tionships were formed further back in timewhen interracial relationships were less accept-ed and they were formed at older ages whenmarriage is more anticipated

Focusing on current sexual relationships thefirst part of our study examines age differencesin the interracial involvement of young adultsduring two different periods It shows howchances of interracial involvement change overthe course of young adulthood Considering therole that the transition to marriage plays in theage decline the study shows the age decline ininterracial involvement before and after themarital status of the relationship is taken intoaccount Residual age differences in interracialinvolvement (ie those that persist once mari-tal status is taken into account) are thought toreflect age period and cohort differences inthe formation of interracial relationships We arenot able to examine the independent influenceof the year and age the relationship began on

interracial involvement because these factors areconfounded in analyses of current relationships

Expanding the scope of the analysis to sex-ual relationships formed in young adulthoodbetween the two periods the second part of thestudy allows us to adjudicate better betweenage and period explanations for the age declineWe examine whether younger individuals aremore likely than older individuals to form inter-racial relationships and whether relationshipsare more likely to be interracial the more recent-ly they were formed We also examine transi-tions between different statuses within sexualrelationships Specifically we compare inter-racial and same-race single relationships interms of the likelihood that they will eventuatein cohabitation or marriage

DDAATTAA AANNDD MMEETTHHOODDSS

Data for our analyses come from the NHSLSa survey administered in 1992 to a nationalprobability sample of 3159 adults between theages of 18 and 59 years5 We limit the NHSLSsample of adults to those 18 to 35 years of ageat the date of the survey because this is the peri-od during which the majority of adults marry forthe first time

Data for our analyses of young adults at theturn of this century come from Add Health aschool-based study of adolescents in grades 7through 12 (Harris et al 2003)6 On the basis ofschool rosters Add Health selected a national-ly representative sample of 12105 students inthese schools to participate in an in-home inter-view in 1994 and 1995 In 2001 and 2002 AddHealth reinterviewed 9130 of the wave one in-home respondents Because most of these



5 NHSLS is a study designed by Edward OLaumann and Robert T Michael with support fromthe NICHD and seven other agencies

6 Add Health is a program project designed by JRichard Udry Peter S Bearman and KathleenMullan Harris and funded by a grant P01-HD31921from the NICHD with cooperative funding from 17other agencies Special acknowledgment is due toRonald R Rindfuss and Barbara Entwisle for assis-tance in the original design Persons interested inobtaining data files from Add Health should contactAdd Health Carolina Population Center 123 WFranklin Street Chapel Hill NC 27516-2524(addhealthuncedu)



respondents were between the ages 18 and 25years at the time of the recent interview welimit the Add Health sample to this age group

Both Add Health and NHSLS asked respon-dents about the race of all their sexual partnersin the previous year and about the status oftheir sexual relationships at the time of the inter-view (ie whether they were in cohabiting rela-tionships or marriages) This allows us tocompare the interracial involvement of youngadults in two different periods and to considerthe role that marriage plays in the age declinein interracial involvement We also are able toevaluate competing explanations for the agedecline in interracial involvement because therecent wave of Add Health collected detailedinformation on sexual relationships since thefirst interview a period of about five years

SSAAMMPPLLEESS OOFF RREELLAATTIIOONNSSHHIIPPSS

As mentioned earlier our analyses for the firstpart of the study are based on the current sex-ual relationships of young adult respondentsfrom NHSLS (n = 1659) and Add Health (n =5189) which include single (eg dating)cohabiting and married relationships Ouranalyses for the second part are based on all sex-ual relationships formed by Add Health respon-dents since 1995 (n = 12195) Relationshipsformed after the age of 18 years are examinedbecause we focus on the age decline in youngadulthood Consequently this study examinesvirtually all the sexual relationships formed byAdd Health respondents in young adulthood

For both parts of the study our unit of analy-sis is relationships rather than individualsRespondents could nominate as many sexualpartners as they had in the preceding year (orprevious five years in the case of Add Healthrespondents)7 Although our emphasis in someanalyses of current relationships minimizes thebias toward individuals with several relation-ships a sizeable number of respondents hadmore than one ongoing relationship at the timeof the interview We exclude relationshipsinvolving partners of the same sex because thenumber of these relationships in NHSLS is

small and because we emphasize the transi-tion to marriage8

MMEEAASSUURREEMMEENNTT OOFF RRAACCEE

Respondents in both surveys were asked abouttheir race (ie white black NativeAmericanAlaskan NativeAmerican IndianAsianPacific Islander) and Hispanic status Inboth surveys respondents who reported morethan one race were asked which one best iden-tified them (see Harris and Sim 2002 for esti-mates of the proportion of Add Healthrespondents who report more than one race) Onthe basis of the race that best describes them aswell as their Hispanic status we divide respon-dents into five mutually exclusive groups whiteblack Hispanic Asian or Native AmericanRespondents who identify themselves asHispanic are defined as Hispanic regardlessof their race We exclude from our sample therelationships of respondents who chose NativeAmerican as their best race because of theirsmall number We classify partners using cate-gories identical to those we use for respondents

DDEEPPEENNDDEENNTT VVAARRIIAABBLLEESS AANNDD MMOODDEELLSS

Because the numbers of respondents havingspecific types of partners are small especiallyin NHSLS we cannot make all the contrasts wewould wish in our dependent variableConsequently our dependent variables simplytake into account whether respondents have adifferent-race partner (eg whether a Hispanicrespondent has a white black Asian or NativeAmerican partner) or a same-race partner

A number of studies on interracial andinterethnic marriage apply log-linear analysesto census data (Fu 2001 Rosenfeld 2001)Because these studies are based on variableswith discrete categories (eg education) thesemethods are ideal In contrast we include sev-eral continuous variables (eg the racial com-position of metropolitan areas) in our models ofinterracial involvement Logistic regressionmodels are more appropriate for our studybecause they can include both discrete and con-



7 A total of 334 NHSLS respondents and 390 AddHealth respondents reported having more than onerelationship at the time of the interview

8 Same-sex cohabiting couples are more likelythan opposite-sex cohabiting couples and marriedpartners to be interracial (Jepson and Jepson 2002)



tinuous variables We estimate multilevel mod-els (Guo and Zhao 2000) to deal with the clus-tering of relationships within primary samplingunits which include National Opinion ResearchCenter (NORC) metropolitan areas for NHSLSrespondents and schools for Add Health respon-dents

We also estimate Cox proportional hazardmodels that predict the timing to a coresiden-tial relationship once sexual relationships areformed (Allison 1995)9 Focusing on respon-dents who form a coresidential relationship wethen use logistic regression models to predictwhether the coresidential relationship beginswith marriage10

IINNDDEEPPEENNDDEENNTT VVAARRIIAABBLLEESS FFOORR MMOODDEELLSS BBAASSEEDD

OONN CCUURRRREENNTT RREELLAATTIIOONNSSHHIIPPSS

AGE To examine how current involvement in aninterracial relationship differs across the peri-od of young adulthood we measure the age ofrespondents at the time of the interview

TYPE OF RELATIONSHIP We distinguish thecurrent relationships of young adults accordingto whether they are single married or cohab-iting relationships Those that are not marriagesor cohabiting relationships fall into the ldquosinglerdquocategory


OONN AALLLL YYOOUUNNGG AADDUULLTT RREELLAATTIIOONNSSHHIIPPSS

AGE AND PERIOD For analyses based on all rela-tionships formed in young adulthood we meas-

ure the age and year the sexual relationship wasformed As suggested earlier the likelihood offorming an interracial relationship may initial-ly increase with age as a consequence of thetransition from school to work but later inter-racial involvement may decrease with agebecause of the increased expectation that therelationship will eventuate in marriage We uselinear and squared measures of age to allow fora nonlinear effect Because we do not knowhow the likelihood of forming an interracialrelationship differs by year the relationshipbegan we use dummy variables to measureperiod effects These variables indicate whetherthe relationship was formed in 1997ndash19981999ndash2000 or 2001ndash2002 using relationshipsformed in 1995ndash1996 as the comparison groupWe do not measure cohort effects because theyare a linear combination of age and periodeffects (Glenn 1976 OrsquoBrien 2000)

INTERRACIAL RELATIONSHIP For models ofthe timing to a coresidential relationship andwhether the coresidential relationship is a mar-riage rather than a cohabitation we include avariable indicating whether the relationship isinterracial

CCOONNTTRROOLL VVAARRIIAABBLLEESS FFOORR AALLLL MMOODDEELLSS

All the models control for several characteris-tics of respondents that differ by age period orcohort and that are tangential to the life courseperspective These include indicators denotingwhether respondents are black Hispanic orAsian whether they are female whether theirmother has less than or more than a high schooldegree and whether they are foreign born11 Wealso take into account the racial composition oflocal areas because opportunities for interracialinvolvement may differ systematically with ageas a consequence of how the primary samplingunits were selected particularly in Add HealthFor young adults from NHSLS we determinethe racial composition of metropolitan areas bymatching 1990 census data to the primary sam-pling unit of each respondent We use the pro-



9 For respondents who form a coresidential rela-tionship we measure the number of months betweenthe formation of the sexual relationship and the for-mation of the coresidential relationship For respon-dents who dissolve a single relationship we measurethe number of months between the formation of thesexual relationship and the dissolution of it Forrespondents who are still in a single relationship atthe time of the interview we measure the number ofmonths between the formation of the sexual rela-tionship and the date of the interview Respondentsin these last two categories have censored observa-tions

10 We do not present the results of Cox models thatpredict the timing of marriage after cohabitationbecause the samples are smaller and more select

11 We use mothersrsquo level of education rather thanthat of respondents because their schooling is pre-sumably completed before the period during whichrespondents form a sexual relationship



portion of individuals in respondentsrsquoStandardMetropolitan Statistical Areas (SMSAs) whoare of a different race as a crude measure of theopportunity for having an interracial relation-ship For young adults from Add Health wedetermine the proportion of individuals in theircounties who are of a different race accordingto contextual data that the Add Health projectproduced using 1990 census data

RREESSUULLTTSS

We first examine how current involvement in aninterracial sexual relationship changes over thecourse of young adulthood As stated earlier oursample from NHSLS includes the sexual rela-tionships of respondents 18 to 35 years of agein 1992 and our sample from Add Healthincludes the sexual relationships of respondents18 to 25 years of age at the time of the recentwave 2001ndash2002 Because Add Health is lon-gitudinal we also are able to include a sampleof the current sexual relationships for respon-

dents who were 12 to 19 years of age in 1995the year of the first wave Combining whitesblacks Hispanics and Asians Figure 1 plots thepercentage of current sexual relationships thatare interracial for each two-year age group

The patterns corresponding to adolescentsand adults in the 1990s suggest that the likeli-hood of having an interracial relationshipdeclines considerably during the course of ado-lescence and young adulthood Among adoles-cents from the first wave of Add Health thepercentage of interracially involved individualsdeclines from 268 percent (among 12- to 13-year-olds) to 142 percent (among 18- to 19-year-olds) For young adults from NHSLSinterracial involvement declines from 124 per-cent (among 20- to 21-year-olds) to 69 per-cent (among 34- to 35-year-olds)

It should be noted that relatively few NHSLSrespondents 18 to 19 years of age have an inter-racial relationship (ie 89 percent) We suspectthat this anomalous statistic may reflect the factthat young adults in this age group are less rep-



Figure 1 Adolescent and Young Adult Relationships that Are Interracial by Age at Interview and SurveyCurrent Relationships of White Black Hispanic and Asian Respondents from the NHSLS and AddHealth

Note Add Health = National Longitudinal Study of Adolescent Health NHSLS = National Health and SocialLife Survey



resentative of the general population becauseNHSLS failed to interview institutionalizedpopulations such as college students in dormsIt also should be kept in mind that the sexualrelationships of adolescents are increasinglyselect at younger ages because fewer adoles-cents are having sex at younger ages (CarverJoyner and Udry 2003) For this reason in sub-sequent analyses we drop adolescent relation-ships from wave one of Add HealthDisregarding the extreme it appears that the agedecline in interracial involvement is roughlylinear

The patterns for young adults from AddHealth that pertain to the beginning of this cen-tury similarly show an age decline in interracialinvolvement For young adults interviewed in2001ndash2002 interracial involvement declinesfrom 197 percent (among 18- to 19-year-olds)to 162 percent (among 24- to 25-year-olds) Acomparison of the lines for young adults fromAdd Health and NHSLS shows that within dif-ferent age groups interracial involvement wasfar more common in the early part of this decadethan in the early to mid 1990s Within each age

group it appears that interracial relationshipsare on the average about five percentage pointshigher in Add Health than in NHSLS

Figure 2 shows the age decline in interracialinvolvement for white black and Hispanicrespondents from NHSLS12 Asian respondentswho have the greatest likelihood of interracialinvolvement are not included in these compar-isons because their numbers in some of the agegroups are small even in Add Health We do notinclude NHSLS respondents 18 to 21 years ofage because of the representational problemmentioned earlier and because the number ofminorities in this age group is small ForHispanics interracial involvement declinesmonotonically from one age group to the nextdecreasing from 375 for 22- to 25-year-olds to313 percent for 30- to 35-year-olds Although



Figure 2 Young Adult Relationships that Are Interracial by Race of Respondent and Age at Interview CurrentSexual Relationships of Respondents from the NHSLS

Note NHSLS = National Health and Social Life Survey

12 Additional results available on the ASR onlinesupplement (httpwwwasanetorgjournalsasr2005toc046html) show for both surveys the proportion ofrespondents from different racial groups who have apartner of a specific raceDelivered by Ingenta to

Mr Matharu (cid 32629) ingenta internal live 231F4 (cid 75000325) University of PennsylvaniaLibrary (cid 80001333)

Wed 26 Oct 2005 211018

the patterns for blacks and whites are less sys-tematic there is a general decline among thesegroups Between the extreme age groups inter-racial involvement drops from 73 to 46 percentfor whites and from 149 to 26 percent forblacks

Figure 3 shows analogous statistics for youngadult respondents from Add Health As in Figure2 an age decline is found for whites blacks andHispanics but the pattern is monotonic onlyfor Hispanics Between the two extreme agegroups (18- to 19-year-olds vs 24- to 25-year-olds) interracial involvement declines from162 to 119 percent among whites from 202to 14 percent among blacks and from 449 to329 percent among Hispanics Both Figures 2and 3 show Hispanics as the most likely to havean interracial relationship followed by blacksand whites As mentioned earlier this orderingpartly reflects differences in the relative sizes ofthe racial groups

To examine the period increase in interracialinvolvement we compare results from Figures

2 and 3 Such an exercise is useful only forwhites and blacks because the sample of respon-dents in NHSLS does not include individualswho had difficulty understanding EnglishPresumably the NHSLS sample includes moreassimilated Hispanics who are expected to havea greater likelihood of interracial involvementFocusing on respondents 22 to 25 years of agein both Figures 2 and 3 it appears that interra-cial involvement increased over time for bothwhites and blacks

Figure 4 shows how interracial involvementdiffers across the three types of relationships forwhites 21 to 25 years of age from both surveysWe focus on these groups to compare directlybetween the different types of relationships andacross the two periods This figure shows thatcohabiting relationships are the most likely andmarriages the least likely to be interracial withsingle relationships falling in the middle It alsosuggests that the increase in interracial involve-ment over time was greatest for cohabiting rela-tionships (ie 71 percent = [156 ndash 91]91)



Figure 3 Young Adult Relationships that Are Interracial by Race of Respondent and Age at Interview CurrentSexual Relationships of Respondents from Add Health

Note Add Health = National Longitudinal Study of Adolescent Health



followed by single relationships (ie 53 percent= [119 ndash78]78) and then by marriages (ie39 percent = [75 ndash 54]54) although the esti-mates are not very precise because of smallsample sizes

Figure 5 shows the age decline for respon-dents from Add Health before and after the typeof relationship is taken into account Amongrespondents 18 to 21 and 22 to 25 years of agerespectively 191 and 16 percent of all rela-tionships are interracial In addition even with-in the three different types of relationships anage decline exists This suggests that the tran-sition to marriage is not the only factor explain-ing the age decline in interracial involvementAge period and cohort differences in the for-mation of sexual relationships also may play arole

And Figure 5 similar to Figure 4 shows thatwithin any given age group cohabiting rela-tionships are the most likely to be interracialmarriages are the least likely and single rela-tionships are in the middle of this range

Considering recent sexual relationships amongadults in NHSLS Laumann et al (1994) foundthis same pattern The finding that cohabitingrelationships are more likely than single rela-tionships to be interracial seems to contradict theassertion that cohabiting relationships requiregreater commitment than single relationships(Blackwell and Lichter 2000 2004) If interra-cial relationships are less likely than same-racerelationships to become cohabiting relation-ships then we would expect the proportion ofinterracial relationships to be smaller amongcohabiting couples than among single couplesHowever these results are based on currentrelationships Figure 6 based on all relation-ships sheds light on this pattern by consideringtransitions to cohabitation and marriage for allAdd Health sexual relationships formed inyoung adulthood

Figure 6 displays the probabilities of differ-ent transitions for both interracial and same-racerelationships using a tree diagram with theprobabilities for interracial relationships in



Figure 4 Relationships that Are Interracial by Type of Relationship and Survey Current Sexual Relationships of21- to 25-year-old White Respondents from the NHSLS and Add Health




parentheses13 These probabilities are based onthe transitions of all young adult sexual rela-tionships formed in the past five years by AddHealth respondents They show that singlerespondents in interracial relationships do notdiffer much from respondents in same-race rela-tionships with respect to their chances of form-ing a coresidential relationship (ie acohabitating relationship or marriage) Theprobability of forming a coresidential relation-ship is 247 (ie 211 + 036) for respondentsin same-race relationships and 215 (ie 200+ 015) for respondents in interracial relation-ships Respondents in interracial relationshipsthen are about 87 percent (ie 870 = 215247)as likely as respondents in same-race relation-ships to coreside with a sexual partner

Whereas respondents in interracial relation-ships differ only slightly from their counter-parts in same-race relationships in theirlikelihood of forming a coresidential relation-ship they differ greatly from their counterpartsin the route by which they begin living togeth-er Respondents in interracial relationships areabout half as likely as their counterparts insame-race relationships to begin their coresi-dential relationship with marriage Amongrespondents who coreside with their sexual part-ners the probability of beginning the relation-ship with marriage is 146 (ie 036[036 +211]) for those in same-race relationships and070 (ie 015[015 + 200]) for those in inter-racial relationships The difference betweensame-race and interracial relationships in thelikelihood of marriage (ie 036 vs 015) isstatistically significant

Interracial cohabiting relationships are aboutthree-fifths (ie 596 = 127213) as likely assame-race cohabiting relationships to becomemarriages For example the probability ofmarrying a partner is 213 for respondents insame-race cohabiting relationships and 127 for



Figure 5 Young Adult Relationships that Are Interracial by Type of Relationship and Age Current SexualRelationships of White Black Hispanic and Asian Respondents from Add Health


13 Additional results available on the ASR onlinesupplement (httpwwwasanetorgjournalsasr2005toc046html) show that these general patterns do notdiffer when statistics are broken down by race ofrespondent and partner



respondents in interracial cohabiting relation-ships This difference is statistically significantThe main point is that while single or cohab-iting respondents in interracial relationshipsare much less likely than their counterparts insame-race-relationships to marry their part-ners14

On the basis of the numbers in the tree dia-gram we can estimate the percentage of dif-ferent current relationship types that areexpected to be interracial at the time of theinterview First we multiply the uncondition-al probabilities in the far right column by thesample sizes of same-race and interracial rela-tionships For instance 318 (ie 2238 0142) interracial relationships are estimated



Figure 6 Probabilities of Different Transitions All Young Adult Sexual Relationships of White Black Hispanicand Asian Respondents from Add Health

Note Statistics for same-race relationships are followed by those for interracial relationships in parentheses N =99572238 Assumes sexual relationship began before cohabitation and marriage which is not the case for lessthan 5 percent of the marriages and less than 1 percent of the cohabiting relationships Add Health = NationalLongitudinal Study of Adolescent Health p lt 05 p lt 01 p lt 001 (two-tailed tests)

14 Interracial relationships differ from same-racerelationships not only in their chances of becomingmarriages but also in their likelihood of dissolvingbefore the recent interview Specifically the proba-bility of dissolving a single relationship is 598 forrespondents in same-race relationships and 643 forrespondents in interracial relationships a difference

that is statistically significant Among relationshipsthat become coresidential relationships howeverinterracial relationships do not differ significantlyfrom same-race relationships with respect to stability



to be ongoing single relationships at the timeof the interview whereas 1543 (ie 9957 0155) interracial relationships are estimatedto fall in this category Therefore 17 percent(ie 17 = 318[318 + 1543]) of ongoing sin-gle relationships are estimated to be interracialAccording to analogous calculations 20 per-cent of ongoing cohabiting relationship and 12percent of ongoing marriages are estimated tobe interracial Similar to the results in Figures4 and 5 these calculations suggest that currentcohabiting relationships are the most likely tobe interracial and that current marriages arethe least likely to be interracial15

MMUULLTTIIVVAARRIIAATTEE RREESSUULLTTSS

Table 1 shows age effects for both samples ofyoung adults based on multilevel logistic regres-sion models of interracial involvement amongthe current sexual relationships of white blackHispanic and Asian respondents16 The firsttwo columns show the odds ratios and logitcoefficients for age in models with control vari-

ables whereas the last two columns show thesesame statistics for models that add an indicatorvariable denoting whether the relationship is amarriage17 Considering the effect of age beforethe marital status of the relationship is taken intoaccount we see that in NHSLS the odds ofinterracial involvement decrease by 53 percentwith each additional year of age (ie [1 ndash 947] 100) and that in Add Health the oddsdecrease by 49 percent with each year of age(ie [1 ndash 951] 100)

As can be seen the age effect is significantat a p value less than 05 until marital status istaken into account It appears that more thantwo-fifths of the age decline in interracialinvolvement in both samples (ie 407 = [054ndash 032]054 for NHSLS and 600 = [050 ndash020]050 for Add Health) is attributable to thefact that the relationships of older individuals aremore likely to be marriages rather than cohab-iting or single relationships Although it is inter-esting from an accounting perspective tounderstand how much of the age decline ininterracial involvement is attributable to thetransition to marriage these models reverse theordering of these events In reality the date ofmarriage follows the date the relationship began



Table 1 Age Effects in Multilevel Logistic Regression Models of Interracial Relationships by Survey CurrentSexual Relationships of NHSLS and Add Health Respondents

Without Marriage Indicator With Marriage Indicator

Survey Odds Ratio Logit Coefficient Odds Ratio Logit Coefficient

NHSLS (N = 1639)mdashAge 947 ndash054 969 ndash032

(018) (020)Add Health (N = 5189)mdashAge 951 ndash050 980 ndash020

(022) (023)

Notes Standard errors are in parentheses Models control for the following characteristics of respondents gen-der race foreign born maternal education and local opportunity Samples include white black Hispanic andAsian respondents The NHSLS sample includes the oversample of Hispanics NHSLS = National Health andSocial Life Survey Add Health = National Longitudinal Study of Adolescent Health p lt 05 p lt 01 p lt 001 (two-tailed tests)

17 Because the effect of age fails to differ signifi-cantly at a p level less than 05 by race in both sam-ples we do not include interaction terms between ageand indicator variables for black Hispanic and AsianAdditional results are available on the ASR onlinesupplement (httpwwwasanetorgjournalsasr2005toc046html)

15 These calculations differ slightly from statisticsbased on the sample of current relationships includ-ed in Figure 5 because the sample of relationshipsformed in young adulthood excludes relationshipsformed in adolescence or before 1995

16 Additional results available on the ASR onlinesupplement (httpwwwasanetorgjournalsasr2005toc046html) show descriptive statistics for the vari-ables in these models in addition to coefficients forbaseline models They also show how the





Still these results suggest that much of the agedecline in interracial involvement is explainedby the shift from single and cohabiting rela-tionships to marriages during the transition toadulthood18

Differences in the age or year that relation-ships were formed also may explain part of theage decline but they are confounded in samplesof current relationships Table 2 estimates thelikelihood of interracial involvement among allyoung adult sexual relationships formed in thepast five years by white black Hispanic andAsian respondents from Add Health It allowsus to examine both age and period effects on theformation of interracial relationships net of

control variables Here we find little support forthe anticipatory mate selection argument thatindividuals are less likely to form an interracialrelationship as they get older because of thegreater anticipation that they will marry Neitherthe squared nor the linear term for age when therelationship began is significant19

The period explanation for the age decline ininterracial involvement appears to be muchmore promising as indicated by the significanteffects of indicator variables for the time whenthe relationship began Relationships areincreasingly likely to be interracial the laterthey are formed in historical time In compari-son with relationships formed in 1995ndash1996

Table 2 Age and Period Effects in Multilevel Logistic Regression Models of Interracial Relationships AllYoung Adult Sexual Relationships of Add Health Respondents

Variables Odds Ratio Logit Coefficient

Age When Relationship Began 1538 431(353)

Age When Relationship Began Squared 990 ndash010(009)

Year Relationship Beganmdash1995ndash1996 mdash mdashmdash1997ndash1998 1265 235

(106)mdash1999ndash2000 1379 321

(107)mdash2001ndash2002 1433 360

(120)Cases (N) 12195

Notes Standard errors are in parentheses Models control for the following characteristics of respondents gen-der race foreign born maternal education and local opportunity Sample includes white black Hispanic andAsian respondents Add Health = National Longitudinal Study of Adolescent Health p lt 05 p lt 01 p lt 001 (two-tailed tests)

19 Although not significant the coefficients for theage at which the relationship began suggest that thelikelihood of forming an interracial relationship withage has an inverted U-shape The age at which thederivative of b1(age) + b2(age2) equals zero isndashb12b2 At this age the probability of having aninterracial relationship is maximized Respondentsare most likely to form an interracial relationshipwhen they are 215 years old (ie 431[2 010])Increases in the formation of interracial relation-ships up to this age may reflect increased opportu-nities for interracial contact that accompany thetransition from school to work whereas decreases inthe formation of these relationships beyond this agemay reflect anticipatory mate selection

18 For models of interracial involvement we cal-culated HuberndashWhite standard errors to deal with thecorrelations of errors for individuals with more thanone relationship Additional results are available onthe ASR online supplement (httpwwwasanetorgjournalsasr2005toc046html) Because the p valuesfor the effects of variables in the models did notchange substantively after we calculated these errorswe present the results of multilevel models that sim-ply take into account the clustering of individualswithin primary sampling units current relationshipsof NHSLS and Add Health respondents of differentage groups differ by type of relationship duration ofrelationship year the relationship began and age atwhich the relationship began





relationships formed in 1997ndash1998 have 23 per-cent higher odds of being interracial The oddsthat relationships formed in 1999ndash2000 and2001ndash2002 will be interracial are respectively32 and 36 percent higher than for relationshipsformed in 1995ndash1996 Thus the formation ofinterracial relationships increases monotoni-cally across the year categories we measure Itis important to note that more recently formedrelationships overrepresent individuals fromlater birth cohorts

Table 3 addresses how interracial relation-ships differ from same-race relationships intheir chances of eventuating in cohabitation ormarriage The first two columns show the haz-ard ratios and estimates for interracial involve-ment based on Cox proportional hazard modelsthat predict the timing to a coresidential rela-tionship For those who form a coresidentialrelationship the last two columns show theodds ratios and logit coefficients for the effectsof interracial involvement based on logisticregression models of whether the coresidentialrelationship was begun with marriage Thesemodels are stratified by the year the relationshipbegan because we find large differences in theformation of interracial relationships accordingto period They include a variable for the age atwhich the relationship began in addition to thecontrol variables

The results in Table 3 show that interracialrelationships are not significantly less likelythan same-race relationships to become cores-idential relationships even if formed in the ear-liest period This finding is at odds with the

assumption that coresidential relationshipsrequire greater investment than single relation-ships (Blackwell and Lichter 2000 2004) Insupport of a winnowing process however inter-racial relationships are significantly less likelythan same-race relationships to become cores-idential relationships via the route of marriageexcept for relationships formed in the mostrecent period Furthermore the patterns in thistable suggest that differences between interra-cial and same-race relationships in terms ofmarriage likelihood are somewhat less pro-nounced for relationships formed later in his-torical time

CCOONNCCLLUUSSIIOONN

Our study tracks and explains changing pat-terns of involvement in interracial sexual rela-tionships during the transition to adulthoodThe results suggest that among current sexualrelationships in two different historical periods(roughly 1990 and 2000) interracial involve-ment declined with age for whites blacks andHispanics The results also suggest that betweenthese two periods current involvement in inter-racial relationships increased more amongcohabiting and single relationships than amongmarriages

Results based on current sexual relationshipssuggest that in both periods the age decline ininterracial involvement reflects the transitionto marriage during young adulthood As evi-dence of this age differences in interracialinvolvement lessen considerably once the mar-

Table 3 Interracial Relationship Effects in Models of Coresidence and Marriage by Year Relationship BeganAll Young Adult Sexual Relationships of Add Health Respondents

Timing to Coresidence Marriage (vs Cohabitation)

Year Interracial Relationship Hazard Ratio Cox Estimate Odds Ratio Logit Coeffificent

1995ndash1996 (N = 1070354) 691 ndash369 139 ndash1970(193) (774)



2001ndash2002 (N = 2733413) 1096 092 566 ndash569(138) (526)

Notes Standard errors are in parentheses Models control for the following characteristics of respondents gen-der race foreign born maternal education local opportunity and age relationship began Sample includes whiteblack Hispanic and Asian respondents Add Health = National Longitudinal Study of Adolescent Health p lt 05 p lt 01 p lt 001 (two-tailed tests)



ital status of the relationship is taken intoaccount Results based on all young adult sex-ual relationships formed between the two peri-ods offer additional evidence that the age declineis a by-product of the transition to marriageConsistent with a winnowing process interra-cial relationships are considerably less likelythan same-race relationships to lead to mar-riage although differences between these rela-tionships in terms of marriage likelihood appearto have lessened in more recent years

We also speculated that the age decline ininterracial involvement might reflect the possi-bility that within periods older individuals areless likely than younger individuals to forminterracial relationships because of their greateranticipation that their relationships will even-tuate in marriage We found no support for theanticipatory mate selection argument Ouranalysis shows that relationships are more like-ly to be interracial the more recently they wereformed suggesting that the age decline mayinstead reflect the increasing acceptance ofinterracial relationships Relationships formedfurther back in historical time overrepresentearlier birth cohorts that may have less tolerantsocial distance attitudes and fewer opportunitiesfor interracial contact

By restricting their focus to marriage previ-ous studies have overlooked the possibility thatincreases in the formation of interracial rela-tionships may have contributed to recent increas-es in cohabitation Whereas individuals areincreasingly likely to form interracial relation-ships the tendency to favor same-race partnersas spouses and different-race partners as room-mates has lessened only slightly This means thata greater proportion of relationships are beingchanneled into cohabitation and correspond-ingly a smaller share of relationships are beingfunneled into marriage But even more impor-tantly previous studies have underestimated theincrease in interracial involvement during recentdecades and they have disregarded the agedecline in interracial involvement

Our study focuses on broader factors impli-cated in the age decline in interracial involve-ment Future studies need to consider whetherthe increased diversity among the young adultpopulation translates into greater opportunitiesfor interracial contact Further research alsoneeds to examine whether the transition fromschool to work suppresses the age decline in

interracial involvement by taking into accountthe institutional contexts of respondents at thetime they formed their relationships We pre-sume that families and friends also affect the for-mation and development of inter racialrelationships and that the strength of theseinfluences changes over the life course This isan issue that future research additionally needsto explore Finally subsequent studies need tomake more refined distinctions in identifyingthe race of respondents and partners Like pre-vious research based on Add Health studiesneed to consider the multiple-race identificationof non-Hispanics (Harris and Sim 2002) therace of Hispanics (Quillian and Campbell 2003)and the ethnicity of Asians and Hispanics (Kingand Harris 2004)

Kara Joyner is Assistant Professor of Policy Analysisand Management at Cornell University Her researchaddresses why individuals form and maintain a vari-ety of relationships including friendships romanticrelationships cohabiting relationships and mar-riages It also concerns how individuals think feeland behave within these relationships Currentlyshe is a Principal Investigator on a subproject TheTiming and Circumstances of the Transition toFatherhood of a program project directed by HElizabeth Peters and funded by the National Instituteof Child Health and Human Development(NICHD)

Grace Kao is Associate Professor of Sociology andthe Director of the Asian American Studies Programat the University of Pennsylvania She is the PrincipalInvestigator (with Joyner as the CoprincipalInvestigator) of the project Interracial Friendship andRomance among Adolescents funded by NationalInstitute of Child Health and Human Development(NICHD) This article is one of several resultant arti-cles from that project Her previous research focusedon race ethnicity and immigrant differences in edu-cational outcomes among adolescents Currentlywith support from the Spencer Foundation and theRussell Sage Foundation she is examining earlychildhood outcomes of immigrant youth

RREEFFEERREENNCCEESS

Aldridge Delores P 1978 ldquoInterracial MarriagesEmpirical and Theoretical Considerationsrdquo Journalof Black Studies 8355ndash68

Allison Paul D 1995 Survival Analysis Using theSAS System A Practical Guide Cary NC SASInstitute

Blackwell Debra L and Daniel T Lichter 2000ldquoMate Selection among Married and CohabitingCouplesrdquo Journal of Family Issues 21275ndash302

mdashmdashmdash 2004 ldquoHomogamy among Dating





Cohabiting and Married Couplesrdquo SociologicalQuarterly 45719ndash37

Blau Peter M 1994 Structural Contexts ofOpportunities Chicago University of ChicagoPress

Blau Peter M Terry C Blum and Joseph Schwartz1982 ldquoHeterogeneity and IntermarriagerdquoAmerican Sociological Review 4745ndash62

Bogardus Emory 1967 A Forty-Year RacialDistance Study Los Angeles University ofSouthern California Press

Bumpass Larry and Hsien-Hen Lu 2000 ldquoTrendsin Cohabitation and Implications for ChildrenrsquosFamily Contexts in the United Statesrdquo PopulationStudies 5429ndash41

Carver Karen Kara Joyner and J Richard Udry2003 ldquoNational Estimates of Adolescent RomanticRelationshipsrdquo Pp 23ndash56 in Adolescent RomanticRelations and Sexual Behavior Theory Researchand Practical Implications edited by P FlorsheimMahwah NJ Lawrence Erlbaum and Associates

Casper Lynne M and Philip M Cohen 2000 ldquoHowDoes the POSSLQ Measure Up HistoricalEstimates of Cohabitationrdquo Demography37237ndash45

Cornell Stephen and Douglas Hartmann 1998Ethnicity and Race Making Identities in aChanging World Thousand Oaks CA Pine ForgePress

Crosnoe Robert 2000 ldquoFriendships in Childhoodand Adolescence The Life Course and NewDirectionsrdquo Social Psychology Quarterly63377ndash91

Elder Jr Glen H 1998 ldquoThe Life Course andHuman Developmentrdquo Pp 939ndash91 in Handbookof Child Psychology edited by Richard M LernerNew York Wiley

Farley Reynolds 1996 The New American RealityNew York Russell Sage

Forste Renata and Koray Tanfer 1996 ldquoSexualExclusivity among Dating Cohabiting andMarried Womenrdquo Journal of Marriage and theFamily 5833ndash47

Fu Vincent Kang 2001 ldquoRacial IntermarriagePairingsrdquo Demography 38147ndash59

Glenn Norval D 1976 ldquoCohort Analystsrsquo FutileQuest Statistical Attempts to Separate Age Periodand Cohort Effectsrdquo American SociologicalReview 41900ndash904

Gordon Milton 1964 Assimilation in AmericanLife New York Oxford University Press

Guo Guang and Hongxin Zhao 2000 ldquoMultilevelModeling for Binary Datardquo Annual Review ofSociology 26441ndash62

Harris David and Hiromi Ono 2004 ldquoHow ManyInterracial Marriages Would There Be If AllGroups Were of Equal Size in All Places A NewLook at National Estimates of InterracialMarriagerdquo Social Science Research 34 236ndash51

Harris David and Jeremiah Joseph Sim 2002 ldquoWhois Multiracial Assessing the Complexity of LivedRacerdquo American Sociological Review 64614ndash27

Harris Kathleen Mullan Francesca Florey JoyceTabor Peter S Bearman Jo Jones and J RichardUdry 2003 The National Longitudinal Study ofAdolescent Health Research Design (Online)Retrieved September 1 2004 (httpwwwcpcunceduprojectsaddhealthdesign)

Heaton Tim B 2002 ldquoFactors Contributing toIncreasing Marital Instability in the United StatesrdquoJournal of Family Issues 23392ndash409

Heaton Tim B and Stan L Albrecht 1996 ldquoTheChanging Pattern of Interracial Marriagerdquo SocialBiology 43203ndash17

Heer David M 1974 ldquoThe Prevalence ofBlackndashWhite Marriage in the United States 1960and 1970rdquo Journal of Marriage and the Family36246ndash58

Hirschman Charles Richard Alba and ReynoldsFarley 2000 ldquoThe Meaning and Measurement ofRace in the US Census Glimpses into the FuturerdquoDemography 37381ndash93

Jacobs Jerry A 1997 ldquoGender Race and EthnicSegregation Between and Within CollegesrdquoDepartment of Sociology University ofPennsylvania Philadelphia PA Unpublishedreport to the Mellon Foundation

Jacobs Jerry A and Teresa G Labov 2002 ldquoGenderDifferentials and Intermarriage among SixteenRace and Ethnic Groupsrdquo Sociological Forum17621ndash46

Jepson Lisa K and Christopher A Jepson 2002 ldquoAnEmpirical Analysis of the Matching Patterns ofSame-Sex and Opposite-Sex CouplesrdquoDemography 39435ndash53

Kalmijn Matthijs 1993 ldquoTrends in BlackWhiteIntermarriagerdquo Social Forces 72119ndash46

mdashmdashmdash 1998 ldquoIntermarriage and HomogamyCauses Patterns and Trendsrdquo Annual Review ofSociology 24395ndash421

King Rosalind Berkowitz and Jenifer Bratter 2004ldquoA Path toward Interracial Marriage WomenrsquosFirst Partners and Husbands across Racial LinesrdquoWorking paper Demographic and BehavioralSciences Branch National Institute of Child Healthand Human Development Bethesda MD

King Rosalind Berkowitz and Kathleen MullanHarris 2004 ldquoRomantic Relationships amongImmigrant Adolescentsrdquo Working paperDemographic and Behavioral Sciences BranchNational Institute of Child Health and HumanDevelopment Bethesda MD

Krysan Maria 2002 ldquoData Update to RacialAttitudes in Americardquo An update and website tocomplement Racial Attitudes in America Trendsand Interpretations Revised Edition Availablehttptiggeruicedu~krysanracialattitudeshtm(Retrieved March 10 2005)







Laumann Edward O John H Gagnon Robert TMichael and Stuart Michaels 1994 The SocialOrganization of Sexuality Chicago University ofChicago Press

Lieberson Stanley and Mary C Waters 1988 ManyStrands Ethnic and Racial Groups inContemporary America New York Russell SageFoundation Census Monograph

Mare Robert D 1991 ldquoFive Decades of EducationalAssortative Matingrdquo American SociologicalReview 5615ndash32

Marsden Peter V 1990 ldquoNetwork DiversitySubstructures and Opportunities for Contactrdquo Pp397ndash410 in Structures of Power and ConstraintPapers in Honor of Peter Blau edited by CCalhoun M Meyer and R S Scott New YorkCambridge

McPherson Miller Lynn Smith-Lovin and James MCook 2001 ldquoBirds of a Feather Homophily inSocial Networksrdquo Annual Review of Sociology27415ndash44

Merton Robert K 1964 Social Theory and SocialStructure London The Free Press of Glencoe

Monahan Thomas P 1970 ldquoAre Interracial MarriagesReally Less Stablerdquo Social Forces 48461ndash73

Moody James 2001 ldquoRace School Integration andFriendship Segregation in Americardquo AmericanJournal of Sociology 107679ndash716

Mouw Ted and Barbara Entwisle Forthcoming ldquoTheEffect of Residential Segregation on InterracialFriendship in Schoolsrdquo American Journal ofSociology

OrsquoBrien Robert M 2000 ldquoAge Period CohortCharacteristic Modelsrdquo Social Science Research29123ndash39

Orfield Gary Mark D Bachmeier David R Jamesand Tamela Eitle 1997 ldquoDeepening Segregationin American Public School A Special Report fromthe Harvard Project on School DesegregationrdquoEquity and Excellence in Education 305ndash24

Park Robert 1924 ldquoThe Concept of SocialDistancerdquo Journal of Applied Sociology 8339ndash44

Qian Zhenchao 1997 ldquoBreaking the Racial BarriersVariations in Interracial Marriage between 1980and 1990rdquo Demography 34263ndash76

Quillian Lincoln and Mary Campbell 2003ldquoBeyond Black and White The Present and Futureof Multiracial Friendship Segregationrdquo AmericanSociological Review 68540ndash66

Raley R Kelly 2001 ldquoIncreasing Fertility inCohabitating Unions Evidence for the SecondDemographic Transition in the United StatesrdquoDemography 3859ndash66

Reardon Sean John T Yun and Tamela McNultyEitle 2000 ldquoThe Changing Structure of School

Segregation Measurement and Evidence ofMultiracial Metropolitan Area School Segregation1989ndash1995rdquo Demography 37351ndash64

Reskin Barbara F Debra B McBrier and Julie AKmec 1999 ldquoThe Determinants andConsequences of Workplace Sex and RaceCompositionrdquo Annual Review of Sociology25335ndash61

Rosenfeld Michael J 2001 ldquoThe Salience of Pan-national Hispanic and Asian Identities in USMarriage Marketsrdquo Demography 38161ndash75

Rytina Steven Peter M Blau Terry Blum andJoseph Schwartz 1988 ldquoInequality andIntermarriage A Paradox of Motive andConstraintrdquo Social Forces 66645ndash75

Schuman Howard Charlotte Steeh Lawrence Boboand Maria Krysan 1997 Racial Attitudes inAmerica Trends and Interpretations CambridgeMA Harvard University Press

Shanahan Michael J 2000 ldquoPathways to Adulthoodin Changing Societies Variability and Mechanismsin Life Course Perspectiverdquo Annual Review ofSociology 26667ndash92

Smock Pamela J 2000 ldquoCohabitation in the UnitedStates An Appraisal of Research ThemesFindings and Implicationsrdquo Annual Review ofSociology 261ndash20

Spickard Paul R 1989 Mixed Blood Intermarriageand Ethnic Identity in the Twentieth-CenturyAmerica Madison The University of WisconsinPress

US Bureau of the Census 1998 Race of Wife byRace of Husband 1960 1970 1980 1991 and1992 (Table 1) Available httpwwwcensusgovpopulationsocdemoraceinterractab1txt(Retrieved March 1 2005)

mdashmdashmdash 2001a Americarsquos Families and LivingArrangements March 2000 Availableh t tp wwwcensus govprod2001pubs p20-537pdf (Retrieved March 1 2005)

mdashmdashmdash 2001b School EnrollmentmdashSocial andEconomic Characteristics of Students October2000 Available httpwwwcensusgovpopulationsocdemoschoolppl-148tab01txt(Retrieved March 1 2005)

mdashmdashmdash 2003 Interracial Married Couples 1980to Present (Table MS-3) Retrieved March 1 2005(httpwwwcensusgovpopulationsocdemohh-famtabMS-3pdf)

Waite Linda J and Kara Joyner 2001 ldquoEmotionalSatisfaction and Physical Pleasure in SexualUnions Time Horizon Sexual Investment andSexual Exclusivityrdquo Journal of Marriage and theFamily



interracial marriage still is relatively uncommonIn 2002 only 29 percent of all marriages wereinterracial (US Bureau of the Census 19982003)2

Since the 1960s men and women have beenpostponing the age at which they first marry andthey are increasingly likely to reach the age of35 years without marrying Furthermore theyare more likely than ever before to form a cohab-iting relationship or accumulate five or more sexpartners during the course of young adulthood(Bumpass and Lu 2000 Casper and Cohen2000 Laumann et al 1994) Although nation-ally representative studies of mate selection andfamily formation have broadened their scope toinclude cohabiting relationships (Raley 2001Smock 2000) they generally have disregardedsingle (eg dating) relationships (for excep-tions see Forste and Tanfer 1996 Waite andJoyner 2001)

The couple of representative studies thatexamine the racial mix of partners in both sin-gle and cohabiting relationships show that indi-viduals in these types of relationships are morelikely than married individuals to have a part-ner of a different race (Blackwell and Lichter2004 Laumann et al 1994) Although innova-tive these studies focus only on current or rel-atively recent relationships in the early or mid1990s Consequently we do not know the extentto which involvement in different types of inter-racial sexual relationships has increased in thepast decade Nor do we know why marriages areless likely than other types of sexual relation-ships to be interracial Researchers speculatethat individuals are increasingly selective withrespect to race (and other traits) as they transi-tion from single relationships to cohabitation tomarriage but they have not been able to exam-ine matching processes fully because of incom-plete data on the histories of sexual relationships(Blackwell and Lichter 2004)

Previous studies have yet to fully theorize orexamine how involvement in interracial rela-tionships changes with age In a study that doesaddress age differences in interracial marriageHeaton and Albrecht (1996) found that in both

1980 and 1990 the likelihood of an individualhaving a spouse of a different race declined sig-nificantly with age3 They argued that the agedecline mainly reflects the fact that youngerindividuals began their marriages in more recentyears when interracial marriage was more com-mon We predict that interracial involvementacross a broader range of sexual relationshipswill show an even greater decline with ageAfter all individuals are increasingly likely tobe married as they get older and marriages areless likely than single and cohabiting relation-ships to be interracial

If interracial marriage is considered to be socritical why is it worth investigating single andcohabiting relationships Like interracial mar-riage these other types of interracial relation-ships are indicators of the social barriersbetween racial groups but they also may con-tribute to the erosion of these barriers by allow-ing individuals to learn about variation withinracial groups and to question racial stereotypes(Kalmijn 1998) Additionally it is critical toconsider the racial mix of partners in single andcohabiting relationships because fertility rateswithin these relationships are increasing (Raley2001) Finally understanding other types ofinterracial relationships helps to enhance ourunderstanding of interracial marriages Kingand Bratter (2004) found that women are morelikely to select husbands across racial lines ifthey select first sex partners across these lines

To foreshadow the age decline we use a lifecourse perspective that highlights how rela-tionships are socially structured by age and his-torical context (Crosnoe 2000 Elder 1998Shanahan 2000) Our analyses are based onsome of the first nationally representative sur-veys to collect detailed information on sexualrelationships the National Longitudinal Studyof Adolescent Health (Add Health) and theNational Health and Social Life Survey(NHSLS)

Focusing on current sexual relationships thefirst part of our study uses data from both sur-veys to examine age differences in interracialinvolvement among young adults in two dif-ferent periods (roughly 1990 and 2000) Turning



2 These census statistics are based on the follow-ing categories of race white black American IndianAsianPacific Islander and other Hispanics are notincluded as a separate category

3 The study by Heaton and Albrecht (1996) exclud-ed marriages that included a Hispanic or Asian part-ner



















































































RREESSUULLTTSS





















Wed 26 Oct 2005 211018























































(022) (023)


















(106)mdash1999ndash2000 1379 321

(107)mdash2001ndash2002 1433 360

(120)Cases (N) 12195





















2001ndash2002 (N = 2733413) 1096 092 566 ndash569(138) (526)








































































































































































RREESSUULLTTSS





















Wed 26 Oct 2005 211018























































(022) (023)


















(106)mdash1999ndash2000 1379 321

(107)mdash2001ndash2002 1433 360

(120)Cases (N) 12195





















2001ndash2002 (N = 2733413) 1096 092 566 ndash569(138) (526)


































































































Wed 26 Oct 2005 211018























































(022) (023)


















(106)mdash1999ndash2000 1379 321

(107)mdash2001ndash2002 1433 360

(120)Cases (N) 12195





















2001ndash2002 (N = 2733413) 1096 092 566 ndash569(138) (526)














































































































































(022) (023)


















(106)mdash1999ndash2000 1379 321

(107)mdash2001ndash2002 1433 360

(120)Cases (N) 12195





















2001ndash2002 (N = 2733413) 1096 092 566 ndash569(138) (526)



































































































(106)mdash1999ndash2000 1379 321

(107)mdash2001ndash2002 1433 360

(120)Cases (N) 12195





















2001ndash2002 (N = 2733413) 1096 092 566 ndash569(138) (526)







































































































2001ndash2002 (N = 2733413) 1096 092 566 ndash569(138) (526)
























































































Interracial Relationships and the Transition to Adulthood

Documents

Transcript of Interracial Relationships and the Transition to Adulthood