Marital Dissolution Among Interracial Couples


  • This article was edited by Velma McBride Murry.

Food and Drug Administration (


Increases in interracial marriage have been interpreted as reflecting reduced social distance among racial and ethnic groups, but little is known about the stability of interracial marriages. Using six panels of Survey of Income and Program Participation (N = 23,139 married couples), we found that interracial marriages are less stable than endogamous marriages, but these findings did not hold up consistently. After controlling for couple characteristics, the risk of divorce or separation among interracial couples was similar to the more-divorce-prone origin group. Although marital dissolution was found to be strongly associated with race or ethnicity, the results failed to provide evidence that interracial marriage per se is associated with an elevated risk of marital dissolution.

Interracial marriage has long been a topic of interest and controversy in American history and has received a great deal of attention in the family research literature (Fu, 2006; Kalmijn, 1991; Tucker & Mitchell-Kernan, 1990; Yancey, 2007). The antimiscegenation laws in the United States, enacted mainly to prevent Black-White interracial marriages, were struck down in a 1967 Supreme Court decision (Sollors, 2000). Since then interracial marriage has increased dramatically from less than 1% in 1970 among all married couples to more than 5% in 2000. Children living in such families have quadrupled to more than 3 million between 1970 and 2000 (Lee & Edmonston, 2005). Such changes have been interpreted as signifying the fading of racial boundaries in U.S. society (Qian & Lichter, 2007) and as indicating immigrant structural assimilation (Alba & Golden, 1986; Gordon, 1964).

Enthusiasm about increases in the prevalence of interracial marriages, however, may be dampened if such marriages are highly likely to break up. Partially because interracial marriage remains a relatively new phenomenon, few studies have assessed the stability of interracial marriages or offered theoretical guidance on this issue. Existing work tends to be dated and focused primarily on Black-White marriages. As a result, little is known about relative stability of such marriages in contemporary American society (Joyner & Kao, 2005). Because the U.S. population has grown increasingly diverse, it is important to update prior research to include interracial marriages involving Asians and Hispanics, especially given that they are more likely to intermarry (with non-Hispanic Whites) than are Blacks (Qian, 1997). Also, interracial marriages involving America’s newest minority groups may operate differently than those involving Blacks because of the high levels of racism in the United States directed specifically toward Blacks, which is likely to stress Black-White marriages. In the present study, we analyze the stability of interracial marriages involving Blacks, Asians, and Hispanics over the period 1990 to 2001 by analyzing data from the Survey of Income and Program Participation (SIPP).

Background and theory

Existing work on marital dissolution among interracial couples focused primarily on divorce within Black-White marriages (Monahan, 1970; Rankin & Maneker, 1987) or specialized populations such as Hawaiians (Fu, 2006; Jones, 1996). Although one study of couples in Iowa found Black-White marriages to be more stable than Black-Black marriages (Monahan), other studies concluded that interracial marriages were less stable in Hawaii (Fu; Jones, 1996) and in the Netherlands (Kalmijn, de Graaf, & Janssen, 2005). Moreover, prior research suggested that the stability of interracial marriages differed by gender. On the basis of a California sample, Rankin and Maneker found that Black men–White women marriages had shorter durations compared to other types of pairings.

Primarily two theoretical frameworks have guided research on the instability of interracial marriages. The first concerns the role of homogamy and the second involves ideas about ethnic convergence of divorce propensities. Consistent with the homogamy perspective, sociologists have long found that people tend to date and marry someone who shares a similar cultural background and social economic status and, in many cases, someone in the same neighborhood, school, or workplace. Among the explanations for homogamy, geographical propinquity and personal preferences were found to be the two underpinning factors (Stevens, 1991). The pool of potential partners is determined by local demographic and geographic composition, and within this pool people generally prefer someone who is similar to them (Kalmijn, 1998; Stevens & Swicegood, 1987).

With respect to marital stability, the basic assumption of the homogamy perspective is that couples with similar characteristics have fewer misunderstandings, less conflict, and enjoy greater support from extended family and friends. Consistent with this idea, endogamous marriages were found to be more stable than those involving couples who were dissimilar on socially significant traits (e.g., education, race or ethnicity, and religion; Bahr, 1981; Jones, 1996; Kalmijn et al., 2005). Kalmijn (1998) found that interracial couples in particular could face group sanctions if racial heterogamy threatened in-group solidarity. Indeed, interracial couples often faced pressures in the forms of strangers’ stares and anger, and even rejection by their own racial groups because of their “betrayal” and nonconforming behavior of “crossing the line” (Billingsley, 1968). This may be particularly salient for Black-White marriages, as Yancey (2007) found that social discrimination against such couples could be especially harsh. The enduring social boundaries specifically between Blacks and Whites became apparent in the continuation of low levels of Black-White intermarriage and higher levels of intermarriage between non-Hispanic Whites, Asians, and Hispanics, especially among those with higher levels of socioeconomic status (Qian & Lichter, 2007). Similarly, in their study of multiracial identification among those with Black, Asian, or Hispanic backgrounds, Lee and Bean (2007) found that those with Black backgrounds more consistently identified as Black and not multiracial (similar to the “one-drop” rule as applied in the past), whereas those with Hispanic and, especially, Asian backgrounds exhibited more flexibility and choice in racial or ethnic identification and were more likely to identify as multiracial. Lee and Bean concluded that these patterns illustrated the salience of the color line that continues to divide Blacks from non-Blacks in U.S. society.

The homogamy perspective predicts that interracial marriages will be less stable than same-race marriages. Thus, Black-White marriages are expected to be more likely to divorce than either Black or White endogamous marriages; similarly, Asian-White marriages are expected to be more likely to divorce than either Asian or White endogamous marriages. The homogamy perspective further leads to the expectation that the stronger the racial boundary of the two groups represented in the couple, the greater the risk of divorce. Thus, Black-White marriages are expected to be at greater risk of divorce than Hispanic-White or Asian-White marriages.

An alternative to the homogamy perspective is the ethnic divorce convergence perspective. Developed by Jones (1996), the ethnic divorce convergence perspective theorizes that divorce propensities of interracial couples are likely to fall between the divorce patterns of the involved racial groups. This contrasts with the homogamy hypothesis, which predicts higher levels of divorce for interracially married couples. On the basis of his empirical tests using Australian and Hawaiian data, Jones (1994, 1996) argued that divorce patterns for mixed marriages reflect the interplay of the divorce cultures of the ethnic groups involved, instead of following the divorce culture of the dominant group. For example, he found that Chinese-White couple divorce rates fell somewhere in between divorce rates of Chinese and White endogamous marriages. Similarly, Kalmijn et al. (2005) reported that interracial marriage divorce rates tended to fall in between the two groups in the Netherlands.

In the United States, Blacks have had higher divorce rates than Whites, whereas Hispanics and Asians have had lower divorce rates (Frisbie, Bean, & Eberstein, 1980; Sweezy & Tiefenthaler, 1996), although there have been important differences among the various Hispanic subgroups, with Puerto Ricans having had the highest divorce rates and Cubans, Mexicans, and Mexican Americans the lowest (Landale & Ogena, 1995). Thus, the ethnic convergence hypothesis would lead to the expectation that Black-White marriages will be less likely to dissolve than Black-Black marriages but more likely than White-White marriages. Similarly, Hispanic-White and Asian-White marriages would be expected to be more likely to dissolve than Hispanic or Asian endogamous marriages but less likely than White endogamous marriages. Jones’s (1996) hypothesis could be viewed especially important for couples involving a combination of immigrants and natives because immigrants have had a tendency toward lower levels of divorce than natives (Frisbie, Opitz, & Kelly, 1985), a difference that has been interpreted as reflecting cultural differences (Oropesa & Landale, 2004). Therefore, according to the ethnic convergence hypothesis, immigrant-native marriages would be expected to have divorce risks that fall between those of immigrant-immigrant and native-native marriages. Also, if Hispanic and Asian interracial marriages are less likely to divorce, this could be because so many of these marriages involve immigrants. After controlling for immigration characteristics, the effects of interracial marriage should diminish for these couples.

To assess the homogamy and ethnic convergence hypotheses, it is important to control for correlated factors. Individual-level socioeconomic and demographic characteristics are associated with interracial marriage and are important predictors of divorce. Age at marriage has been one of the most consistent key determinants for predicting marital stability (South, 1995), which also have differed by racial group (Phillips & Sweeney, 2006). Also, people with medium levels of education (Carter & Glick, 1970) and women with higher incomes (Ruggles, 1997) have been more likely to divorce or separate than others. Finally, although having young child(ren) has been shown to increase marital stability, this effect often decreased as the child(ren) grew older (Cherlin, 1977).

In addition to the socioeconomic and demographic characteristics of individuals, it is critical to control for couple-level characteristics. The homogamy perspective stresses that partner differences in any socially significant characteristics—not just race—may increase the risk of divorce, and spouses in interracial marriages may differ on multiple characteristics. For example, Tucker and Mitchell-Kernan (1990) found that the age gap was larger for interracially married couples than other couples. Partners in interracial couples may also differ with respect to nativity and citizenship. Interracial marriages between immigrants and U.S.-born natives may be at greater risk of divorce because of partner differences in their reasons for entering the relationship. Kalmijn et al. (2005) found that larger cultural differences between the husband and wife increased the risk of divorce. In addition, marriage to U.S. citizens may serve as a legal means to immigrate for many foreigners. Such marriages may be motivated by the desire to obtain U.S. citizenship rather than love or companionship, as evidenced in many cases in France (Neyrand & M’Sili, 1998) and the Netherlands (Kalmijn et al.).

Finally, group-level characteristics, such as marriage cohort, region of residence, religion, and women’s changing status, may be associated with divorce or separation (Trent & South, 1989). For example, interracial marriage has been more prevalent in the West than other parts of the country (Tucker & Mitchell-Kernan, 1990), and marital instability has been more common in the West than other regions, although this relationship has varied by race (Sweeney & Phillips, 2004) and has weakened over the years (Castro Martin & Bumpass, 1989).

To summarize, the homogamy perspective leads to the following hypotheses:

Hypothesis 1. Interracial marriages will be less stable than similar endogamous marriages.

Hypothesis 2. Black-White couples will have higher risks of marital dissolution than other interracial couples.

Hypothesis 3. Interracial couples will have higher risks of marital dissolution than endogamous couples among both of the respective origin groups.

On the other hand, the ethnic convergence hypothesis leads to expectations of the following:

Hypothesis 4: Interracial couples have risks of marital dissolution that fall between those of the respective origin groups.

Hypothesis 5: Among Hispanics and Asians, differences in the risk of marital dissolution between interracial and endogamous couples will be partially explained by differences in nativity and citizenship.



Six panels (1990, 1991, 1992, 1993, 1996, and 2001) of the SIPP were pooled in order to study marital dissolution patterns among interracially married immigrants and natives. The SIPP sample is a multistage-stratified sample of the U.S. civilian noninstitutionalized population with sample sizes ranging from approximately 14,000 to 36,700 interviewed households (U.S. Census Bureau, n.d.). Each of the six SIPP panels includes a separate, independent sample that was interviewed every 4 months for roughly 3 to 4 years. For example, the 1990 panel includes individuals who were interviewed up to eight times over a period of 32 months between 1990 and 1992, and the 1991 Panel includes an entirely new sample interviewed up to eight times over a period of 32 months between 1991 and 1993. The respondents in the 1992, 1993, 1996, and 2001 panels were interviewed every 4 months over 40, 36, 48, and 36 months, respectively. Although the SIPP was not designed to study the same marriage cohorts and it only followed respondents for 3 to 4 years, it provides a unique snapshot of the marital stability of interracially married couples. The advantages of using the SIPP are apparent. First, it followed individuals (and married couples) over time even if they left their original households and formed new ones. Secondly the SIPP included time-varying information on marital status as well as standard social, demographic, and economic variables (these questions were asked at every interview every 4 months). Third, it included a retrospective marriage and migration history for all adult household members.

By combining six SIPP panels, we amassed a sufficiently large sample to examine the stability of interracial marriages separately for couples of various racial/ethnic group combinations. The data used for this paper are mainly couple-level prospective data, which followed couples over time during the 3 to 4 year study period until they divorced or separated, dropped out, or were censored. The analytical sample was first restricted to 29,171 couples with at least one member who was between the ages of 18 to 44 and in a first marriage at the beginning of the SIPP panel or were married for the first time during the ongoing waves of the panel. Because of limitations in sample size, Native American endogamous marriages (n= 91) and White-Native American marriages (n= 232) were excluded. We further restricted our sample by removing 5,709 couples on the basis of censoring or missing or invalid values for the time or any of our explanatory variables; thus, the final sample contains 23,139 couples.

Key measures

The dependent variable was the dissolution of marriage by either divorce or separation for all couples. The timing of marital dissolution was determined prospectively by observing year and month in which a married respondent was coded as having changed marital status to divorced or separated. Widowed respondents were censored at the time they were no longer married (or were dropped from the sample). First marriage was defined as an ongoing first marriage at the beginning of the SIPP panel or transition from never married to married for the first time during the SIPP panel. The duration of marriage (measured in days) was obtained by examining the difference between the date of first marriage and the date of marital dissolution. For presentation, the descriptive statistics on marriage duration were reported in years instead of days. Marital duration is implicitly embedded in all of our Cox Proportional Hazards models (duration is not included as an independent variable because the partial likelihood function takes into account the ordering of failure times; Box-Steffensmeier & Jones, 2004). Interracial couples were identified as those married to a person of a different race or ethnicity. We opted to use broadly defined racial and ethnic categories: non-Hispanic White, non-Hispanic Black, Asian, Hispanic, and other minority. We refer to non-Hispanic White and non-Hispanic Black with the shortened terms “White” and “Black” throughout the remainder of the paper. Marriage types were later further classified by gender (e.g., White husband-Black wife, Black husband-White wife).

Control variables

A set of six dummy variables was included in the models to indicate the time period (in 5-year intervals) the couple got married (e.g., before 1970, 1970 – 1974, …, 1990 – 1994, and 1995 or later). Region of residence at the time of the first interview was classified according to the U.S. Census Bureau standard (Midwest, Northeast, South, and West). Because age at first marriage showed clear nonlinearity in our preliminary analysis, we opted to code it as a series of dummy variables. The age difference between the spouses was categorized in the same fashion as by Phillips and Sweeney (2006): husband more than 5 years older than the wife, less than 2 years younger to 5 years older than the wife, or more than 2 years younger than the wife. To measure educational heterogamy, we distinguished between couples for whom the husband had more or less education than his wife. For couples who have same educational levels, we distinguished them into four categories: less than high school, high school, some college, and college or above. We controlled for the wife’s income (logged) by summing across income from the previous 4 months at the time of the first interview. Because some respondents have a true zero for their income, log (income + 1) was used instead of log (income). We reported the unlogged mean income (summed over 4 months) in our descriptive statistics. We also controlled for the number of preschool-age (0 to 4 years old) children living with the couple and both the citizenship and nativity status of the couple. Specifically, we distinguished among couples in which (a) both spouses were foreign born and at least one was a noncitizen, (b) both spouses were foreign-born naturalized citizens, (c) one spouse was a noncitizen but the other was native, (d) one spouse was a naturalized citizen and the other was native, and (e) both spouses were native (born in the United States).

Data analysis

Cox Proportional Hazards models were used to estimate the relationship between interracial marriage and the hazard of dissolution. All independent variables were non-time-dependent covariates. Cox Proportional Hazards models assume that the underlying hazard rate (rather than survival time) is a function of the independent variables (covariates) and is thus proportional to the hazard function for the baseline category (Allison, 1995). The proportional hazards models are robust and make no parametric assumptions concerning the nature or shape of the underlying survival distribution (Allison).


Descriptive results

Table 1 presents descriptive statistics of all the couples that are included in the final analytical sample (N= 23,139). Approximately 2,059 couples (8.9% of the sample) divorced or separated during the time interval they were followed in the SIPP. The majority (93.5%) of the couples in our sample were endogamous, including 77.4% White-White, 6.4% Black-Black, 7% Hispanic-Hispanic, and 2.7% Asian-Asian couples. The remaining 6.5% of couples were interracially married (including 1% White-Black, 3.5% White-Hispanic, 1.4% White-Asian pairings, and 0.6% of all types of minority-minority marriages combined). Consistent with prior studies (e.g., Qian, 1997), there are distinct racial or ethnic differences in being in an interracial marriage (results not shown). Blacks are substantially less likely than Hispanics or Asians to have a White spouse (10.1% vs. 23.5% and 24.6%, respectively).

Table 1. Potential Predictors of Marital Dissolution by Marriage Type: Descriptive Statistics
VariablesAll CouplesEndogamous CouplesInterracial Couples
(N= 23,139)(n= 21,547)(n= 1,592)
  1.  Note: Means are weighted. Data are from the pooled 1990, 1991, 1992, 1993, 1996, and 2001 Survey of Income and Program Participation panels. Couples in a first marriage at the beginning of the SIPP panel and couples who married during the ongoing waves of the panel are included. Marriages involving Native Americans are not included. H = husband, W = wife.

  2.  aRange between 1 and 32 years. bWhite refers to Non-Hispanic White.cBlack refers to Non-Hispanic Black.dIncome at first spell: income summed over 4 months prior to the first interview, range between $0 and $155,684.eRange between 0 to 5 children.

Marriage duration (years)a12.5006.95012.6197.00010.4495.914
Endogamous couples
 White-Whiteb.774 .828 
 Black-Blackc.064 .069 
 Hispanic-Hispanic.070 .074 
 Asian-Asian.027 .029 
Interracial couples
 All.065 1.000 
 White H-Black W.004 .067 
 Black H-White W.006 .091 
 White H-Hispanic W.018 .281 
 Hispanic H-White W.017 .265 
 White H-Asian W.009 .136 
 Asian H-White W.005 .069 
 Minority-Minority.006 .092 
Marriage cohort
 Before 1970.063 .066 .014 
 1970 - 1974.123 .127 .066 
 1975 - 1979.162 .166 .111 
 1980 - 1984.225 .226 .213 
 1985 - 1989.273 .269 .339 
 1990 - 1994.139 .134 .213 
 1995 or later.015 .012 .045 
Region of residence
 Midwest.264 .271 .159 
 Northeast.184 .187 .138 
 South.350 .353 .303 
 West.203 .189 .401 
Wife’s age at first marriage
 < 20 years.241 .247 .151 
 20 – 22 years.270 .273 .232 
 23 – 26 years.242 .241 .253 
 27 – 30 years.126 .124 .153 
 >30 years.122 .116 .211 
Age categories
 H > 5 years older than W.185 .183 .214 
 H’s w/in −2 to 5 years of W.732 .741 .607 
 H > 2 years younger than W.083 .076 .179 
Education categories
 H more educated than W.187 .184 .236 
 H less educated than W.174 .173 .191 
 Both less than high school.050 .051 .041 
 Both high school.204 .207 .157 
 Both some college.020 .018 .047 
 Both college.365 .368 .328 
Income at first spelld4,955.3506,136.1904,896.1005,990.4605,807.2707,784.820
Number of preschool-aged childrene.4280.676.4250.674.4740.692
 Both natives.859 .877 .605 
 Both foreign born, at least one is noncitizen.059 .060 .035 
 Both foreign born & citizens.020 .021 .014 
 Noncitizen-native.030 .020 .177 
 Naturalized citizen-native.032 .022 .168 

Table 1 highlights compositional differences between interracial and endogamous marriages. For example, compared with endogamous couples, interracial couples married in more recent time periods were more likely to live in the West and less likely in the Midwest, married on average at older ages (wives), had higher incomes and larger spousal differences in age and education, and were much more likely to involve a combination of foreign-born and native-born spouses. Over one third of interracial couples (34.5%) involved a foreign-born person married to a U.S. native compared with just 4.2% of endogamous couples.

As the focus of our analysis is on marital dissolution, we first examined the observed differentials in marital dissolution without considering the possibly confounding factors associated with both interracial marriage and marital instability during the observation period (results not shown). Consistent with the first homogamy hypothesis, interracial marriages are less stable: 13.7% of interracial couples compared with 9.9% of endogamous couples broke up during their SIPP panel. The descriptive results also confirm the second homogamy hypothesis in which mixed-race couples involving the most socially distant groups (e.g., Blacks and Whites) were most likely to break up: Nearly 20% of Black-White couples divorced or separated compared with 13.5% of Hispanic-White couples and 8.4% of Asian-White couples. Furthermore, consistent with the third homogamy hypothesis, both White-Black and Hispanic-White couples were more likely to divorce or separate than endogamous couples from either of the origin groups (10% of White-White, 16% of Black-Black, and 9% among Hispanic endogamous couples). For Asians, however, the results were consistent with the ethnic convergence hypothesis (Hypothesis 4). Roughly 8.3% of Asian-White couples separated or divorced, a level that falls between the relatively high rates for White couples and the relatively low rates among Asian couples (1.4%). The descriptive results thus suggest that interracial couples, especially those involving Blacks and Hispanics, are more likely to divorce or separate than same-race couples. This may be a consequence of potential problems facing interracial couples including stress, social disapproval, and cultural differences. An alternative explanation, however, is that interracial couples differ from endogamous couples in important ways that elevate the risk of divorce (such as greater age and education differences between spouses). To test this idea, we turn next to the multivariate hazards models.

Multivariate results

Table 2 presents the results of the Cox Proportional Hazards models of marital dissolution (hazard ratios are displayed). Model 1 includes an indicator of all types of interracial marriages without any controls (Model 1 does not exactly replicate the descriptive results in Table 2 because it conditions the hazard on the duration of marriage). Indeed, interracial marriages are less stable. The risk of marital dissolution among mixed marriages is about 1.21 times that of (or 21% higher than) nonmixed endogamous marriages (Table 2, Model 1), and this did not change after adding controls for marriage cohort and region of residence (Model 2). When we added other potential marital dissolution risk factors in Model 3, the hazard ratio associated with mixed marriage declined by 25% to 1.15, and dropped in significance (p < .01 to p < .10). In general, younger age of first marriage, age and educational differences among the spouses (particularly when the husband is more than 2 years younger or less educated than the wife), lower levels of education (less than college), lower income, and having no or fewer young children were significantly associated with marital instability. Interracial couples tend to have higher incomes and older ages at marriage (both of which are associated with lower rates of dissolution), so these characteristics cannot explain their higher levels of divorce or separation. Although, mixed marriages are also more likely to involve larger differences in age and education between spouses (consistent with the first homogamy hypothesis), which could partially explain their higher risks of marital dissolution.

Table 2. Hazard Ratios From Cox Proportional Hazards Models of Marital Dissolution (N =23,139)
 Model 1Model 2Model 3Model 4
  1.  Note: Reference categories are shown in parentheses.

  2.  p < .10.*p < .05.**p < .01.***p < .001.

Interracial couple1.21**1.21**1.151.15
Marriage cohort
 (Before 1970)
 1970 – 1974 1.411.361.38
 1975 – 1979 1.371.241.25
 1980 – 1984 1.311.131.15
 1985 – 1989 1.301.091.11
 1990 – 1994
 1995 or later
Region of residence
 Northeast .77***.82**.85*
 South 1.18**1.14*1.15**
Wife’s age at first marriage
 (< 20 years)
 20 – 22 years .69***.70***
 23 – 26 years .59***.60***
 27 – 30 years .53***.54***
 >30 years .45***.46***
Age categories
 (Husband > 5 years older than wife)
 Husband’s age w/in −2 to 5 years of wife .78***.77***
 Husband > 2 years younger than wife 1.30***1.28***
Education categories
 (Husband more educated than wife)
 Both less than high school 1.021.22
 Both high school 1.101.09
 Both some college .96.93
 Both college .63***.62***
 Husband less educated than wife 1.14*1.13*
Log of income at 1st spell 1.04***1.03***
Number of preschool-age children .65***.66***
 (Both natives)
 Both foreign born, at least one noncitizen .43***
 Both foreign born & citizens .48***
 Noncitizen-native .87
 Naturalized citizen-native .94

We next tested the idea that spousal differences in nativity or citizenship status may explain the higher risk of marital dissolution among mixed-race couples. As shown in Model 4, the risk of divorce was significantly lower for foreign-born couples (both spouses foreign born) than native-born couples, whereas mixed-status couples (foreign-born/native pairings) were not significantly different from couples involving two natives. Unexpectedly, however, the addition of controls for nativity/citizenship status did not alter the hazard ratio associated with interracial marriage.

Thus far, the results support the first homogamy hypothesis, though the support was rather weak. Interracial marriage was positively associated with marital dissolution net of couple characteristics, but this relationship was only marginally significant (p < .10). To test the remaining hypotheses, it is necessary to examine the risk of marital dissolution separately across racial and ethnic groups. We therefore reestimated the Cox models shown in Table 2, this time breaking apart the single indicator of interracial marriage into multiple race combinations (upper panel of Table 3). We presented the hazard ratios for race or ethnicity only, although the full models are available to interested readers upon request.

Table 3. Hazard Ratios From Cox Proportional Hazards Models of Marital Dissolution by Race and Gender of the Couple (N =23,139)
Couple TypeModel 1Model 2Model 3Model 4
  1.  Note: Controls were added to the models as in Table 2. Reference categories are shown in parentheses. H = husband, W = wife.

  2.  p < .10, *p < .05, **p < .01, ***p < .001 (significantly different from non-Hispanic White endogamous couples); bp < .10, ap < .05, aap < .01, aaap < .001 (significantly different from minority endogamous couples in the same group).

Other mixed-race couples1.
White H-Black W1.58*1.57*1.441.44
Black H-White W1.85***1.88***1.63**1.62**
White H-Hispanic W1.111.081.05a1.06
Hispanic H-White W1.33*aa1.28aa1.18aa1.19
White H-Asian W.77aaa.74aaa.74aaa.77aa
Asian H-White W.67aa.67aaa.77aa.79a
Other mixed-race couples1.

When we examine the instability of interracial marriages by race or ethnicity in Table 3, the results generally reveal patterns that are more consistent with the ethnic convergence than the homogamy hypothesis. Nevertheless, the results were consistent with the second homogamy hypothesis in that the risk of marital dissolution was highest among Black-White couples, followed by Hispanic-White, minority-minority couples, and finally, Asian-White couples. This ordering was retained across all models, although only Black-White couples had significantly greater hazards of dissolution than White endogamous couples when all controls were included in the model (Model 4, Table 3).

Overall, neither the descriptive nor the multivariate results provide strong support for the third homogamy hypothesis that interracial couples would be less stable than endogamous marriages from each of the origin groups. Across all four models in Table 3, the hazard of dissolution for Black-White couples was significantly higher than White-White couples but not significantly higher than Black-Black couples. Among couples involving Hispanics, the risk of marital dissolution for Hispanic-White couples was 22% greater as compared to White-White couples. Though, when we controlled for couple characteristics in Model 3, the difference between Hispanic-White and White-White couples became insignificant. Also, after controlling for nativity and citizenship in Model 4, the difference between Hispanic-White and Hispanic-Hispanic couples also became insignificant. Thus, only Black-White couples were more likely to break up than otherwise similar White-White couples, but their risk of dissolution was no different from that of Black-Black couples.

Finally, the results provided some weak support for the ethnic convergence hypotheses. Among Asians, the hazard of divorce or separation for interracial couples fell between that of Asian and White endogamous couples, but the difference from White couples was not significant, thus failing to fully support Hypothesis 4. We had also hypothesized that nativity and citizenship between spouses of Hispanic and Asian interracial couples may help explain their higher risks of marital dissolution (Hypothesis 5). This idea was not fully supported because interracial marriages involving Hispanics or Asians did not experience elevated hazards of dissolution (so there were no significant differences to explain). Nevertheless, nativity and citizenship did help explain the relatively low risks of instability among Hispanic and Asian endogamous couples. When we added controls for nativity and citizenship in Model 4, the hazards for Hispanic and Asian endogamous couples increased, thereby narrowing the difference from both White couples and interracial couples. In fact, the difference between Hispanic-White and Hispanic-Hispanic couples became insignificant after controlling for citizenship and nativity in Model 4.

As a final step, we investigated whether the findings varied by gender combination of interracial couples. We repeated the models by distinguishing among various race or ethnicity and gender combinations as shown in the lower panel of Table 3 (hazard ratios only). Overall, interracial marriages involving a minority husband and White wife were less stable than other types of interracial marriages. Among them, Black husband-White wife and Hispanic husband-White wife couples were particularly likely to break up (Model 1). Nevertheless, the results were consistent with the earlier findings showing little to weak support for the third homogamy hypothesis. After all controls were added, Black husband-White wife couples showed higher risks of dissolution than White couples but similar (and not higher) risks as Black couples. Among Hispanic-White couples, Hispanic husband-White wife were no more likely to dissolve than White or Hispanic endogamous couples.


Marital dissolution of interracial marriage serves as an important indicator of the salience and persistence of racial and ethnic boundaries in the United States. Yet the majority of the current studies on interracial marriage have focused on the prevalence of such unions instead of their dissolution, and those that have assessed the instability of interracial marriages tend to be dated and have focused solely on Black-White marriages or on interracial marriage in a particular region. The contribution of this study is that it examines the instability of interracial marriage among Whites, Blacks, Hispanics, and Asians in contemporary American society, an era marked by increasing diversity and increasing prevalence of interracial marriage. Overall, although marital dissolution was found to be strongly associated with race or ethnicity, the results failed to provide evidence that interracial marriage per se is associated with an elevated risk of marital dissolution. Some have argued that the increasing prevalence of interracial marriage may be associated with the reduction of social distance across groups. The findings described here showing little difference between interracial and endogamous marriages in the risk of divorce or separation lends further credence to this (albeit cautiously) optimistic point of view.

Our first hypothesis (consistent with the “homogamy” perspective) predicts that when couples are dissimilar in racial or ethnic background, their risk of divorce or separation is higher. Our results do show that, on the whole, interracial marriages are less stable than endogamous marriages, even after controlling for couple characteristics. Nonetheless, the analysis compared all interracial marriages with all endogamous marriages and did not take into account group variations in the tendency to divorce or separate. When we divided the results by race and ethnicity, the results were only partially consistent with the homogamy perspective. As we expected, racial or ethnic differences appeared in the risk of divorce or separation. Mixed marriages involving Blacks were the least stable followed by Hispanics, whereas mixed marriages involving Asians were even more stable than endogamous White marriages. In addition, Black husband–White wife pairings were found to be the least stable of all marriage types. One plausible interpretation of these results is that they reflect persistent racism and distrust directed toward Blacks, particularly Black men in the United States. The qualitative findings from Yancey (2007) indicated that Whites who married Blacks experienced more firsthand racism as compared to Whites who married other non-Black minorities. Specifically, White women reported encountering more racial incidents with their Black husbands (e.g., inferior service, racial profiling, and racism against their children) and more hostilities from families and cohorts as compared to other interracial pairings (Yancey). Research in communication and cultural studies echoed Yancey’s findings and found that the above-mentioned social pressures tend to increase social isolation of Black-White unions, especially from the White community, and consequently negatively impact the survival of these marriages (Hibbler & Shinew, 2002; Porterfield, 1982).

As suggested by the results that compare interracial couples with endogamous couples within racial and ethnic groups, the racial or ethnic patterns in the risk of dissolution appear to reflect broad racial and ethnic differences, which may in turn be associated with a number of factors, including discrimination, but are not specifically associated with interracial marriage itself. Specifically, in the descriptive results, Black-White and Hispanic-White marriages appeared to be less stable as compared to Black or Hispanic endogamous marriages, respectively (consistent with Hypothesis 3). These patterns, however, failed to appear in any of the multivariate models for Blacks and, in the case of Hispanics, were attenuated and did not reach statistical significance once couple-level characteristics were controlled. Thus, the third homogamy hypothesis was not supported. Rather, the most consistent result was that the risks of divorce for interracial couples for all combinations (Black-White, Hispanic-White, and Asian-White) were not significantly different from those of the higher-risk origin group. Kalmijn et al. (2005) found a similar pattern in the Netherlands in which the risk of divorce for mixed marriages seemed to be driven more by the divorce-prone group of the couple, a pattern they argue to be consistent with “a weak form of a heterogamy effect” (p. 83). More consistent with the ethnic convergence than the homogamy perspective, this idea makes some sense assuming that it takes both partners to make a marriage work. Furthermore, the results concerning Asians and, to some degree, Hispanics offer some additional support for the ethnic convergence perspective. The risk of divorce for Asian-White couples fell between that of Asian and White couples (significant in the descriptive results but not in the multivariate models), and the differences between Asian and Asian-White couples and between Hispanic and Hispanic-White couples narrowed once citizenship and nativity were added to the model.

The research presented here has several limitations. Although the SIPP data offer important advantages over alternative data, they provide only a snapshot of marriages for a 3- to 4-year time period. Even after pooling six SIPP panels together, the number of interracial couples was small, which may have contributed to the insignificant findings. In addition, because of limitations of the SIPP, we were unable to control for some potentially important measures (such as cohabitation history) and macrolevel forces (institutionalized discrimination and racism). Future research on the stability of interracial marriage would benefit from data that permit larger numbers of interracial couples to be followed over a longer period of time. Because many Hispanic and Asian families are still relatively new to the United States, the generational effect may be more apparent in the long run. Also, qualitative data and mixed methodology could help shed light on the complexity of this issue that quantitative data usually cannot easily answer. Finally, better theoretical frameworks need to be developed to guide the explanation of interracial marital stability. An example concerns the relationship between race and gender combinations and marital stability. In our study, the effects of certain racial or ethnic combinations were similar for both men and women once controls were introduced into the models (e.g., among Asians and Hispanics). Still, Black husband-White wife marriages tended to be less stable than White husband-Black wife marriages. This may reflect differences in the social acceptability of certain combinations, but the precise reasons remain unclear. As interracial marriages and multiracial children become more commonplace, expanding research on the stability of interracial marriage becomes vital, not only for the children born into those marriages but also because the success and stability of such marriages provides insight into the enduring or waning social rigidity of racial boundaries in the United States.


We thank Dr. Zhenchao Qian for providing many helpful suggestions and comments on earlier drafts of this paper.