Including black-white couples in the study of residential stratification accentuates gendered power disparities within couples that favor men over women, which allows for the analysis of whether the race of male partners in black-white couples is associated with the racial and ethnic composition of their neighborhoods. I investigate this by combining longitudinal data between 1985 and 2015 from the Panel Study of Income Dynamics linked to neighborhood- and metropolitan-level data compiled from four censuses. Using these data, I assess the mobility of black male–white female and white male–black female couples out of and into neighborhoods defined respectively by their levels of whites, blacks, and ethnoracial diversity. My results show that the race of the male partner in black-white couples tends to align with the racial and ethnic composition of the neighborhoods where these couples reside. This finding highlights that the racial hierarchy within the United States affects the residential mobility and attainment of black-white couples, but its influence is conditioned by the race and gender composition of these couples.
Throughout U.S. history, legal and extralegal proscriptions have surrounded black-white coupling (Romano 2003:39–42, 128). However, with the end of anti-miscegenation laws in 1967 with the U.S. Supreme Court case Loving v. Virginia, a broader softening of public sentiment toward black-white unions has emerged (Bialik 2017). Although contemporarily, black-white marriages account for 8 % of a growing population of U.S. mixed-race couples (U.S. Census Bureau 2015), scholars have argued that black-white couples are an indicator of racial group relations given that interracial coupling is the antithesis of interracial group contention (Fryer 2007).
The increased representation of black-white couples has coincided with greater scholarly interest in investigating their residential location (Wright et al. 2003), leading a small number of scholars to explore how these couples manage in existing systems of residential stratification that have separated blacks and whites (Gabriel 2016; Holloway et al. 2005; Wright et al. 2011). This past research on the residential attainment of black-white couples indicates that these couples are often found in racially and ethnically diverse neighborhoods (Gabriel 2016; Holloway et al. 2005; Wright et al. 2011). The residential attainment patterns of black-white couples have also been shown to decrease broader levels of black-white residential segregation (Ellis et al. 2012).
Building on the work of Ellis et al. (2006), who posited that the race and gender combination of mixed-race couples might influence their residential location, Wright et al. (2013) observed that the race and gender combination of black-white couples is salient in determining the racial and ethnic composition of these couples’ neighborhoods. Their findings demonstrated that black-white couples with white male partners tend to reside in neighborhoods with higher shares of whites and lower concentrations of blacks than when the male partner is black. Their study using 2000 U.S. Census confidential data raises an important question as to the residential mobility and attainment of black male–white female and white male–black female couples. Because of the limits of cross-sectional data, they were unable to account for the potentially disparate residential mobility and attainment processes associated with black-white couples with black or white male partners. Assessing the residential mobility and attainment of these couples is important given that each type of couple might demonstrate a unique mobility pattern because, historically, black male–white female and white male–black female couples experienced drastically different societal treatments. For instance, during the Jim Crow era, many black men who were perceived to make advances toward white women faced brutal violence (Fiemster 2011:92–93). Conversely, white men who sexually assaulted black women functioned with virtual impunity (Romano 2003:236). Although currently less severe, these two different societal reactions to individuals crossing the color line potentially linger in the contemporary milieu within which these black-white couples function, possibly influencing their propensity to migrate given the racial and ethnic composition of their neighborhoods and, once they decide to move, the racial and ethnic composition of the neighborhoods they select.
Another important issue that extends from the research of Wright et al. (2013) is that although the opportunities for couple formation have grown (Kalmijn and Flap 2001), past research has shown that neighborhoods are a place where couples meet (Houston et al. 2005). However, cross-sectional data are limited in that a nontrivial number of black-white couples with black or white male partners might have formed in their respective neighborhoods instead of migrating to those places. The use of longitudinal data that track the residential location of couples allow, to some extent, a greater ability to measure intentionality in the housing search and acquisition process compared with cross-sectional approaches. Moreover, leveraging longitudinal data for the analysis of black-white couples with black or white male partners allows for the assessment of the effect of household socioeconomic resources on the racial and ethnic composition of the neighborhoods these couples migrate to—a key theoretical question related to the potential assimilation of these couples that has yet to be investigated. Also clear from past studies is that the ethnoracial and socioeconomic status of neighborhoods are linked: neighborhoods with higher white concentrations tend to have substantially higher levels of socioeconomic status as compared with black and diverse neighborhoods (e.g., Massey and Tannen 2016). Consequently, when assessing the residential mobility of couples out of neighborhoods defined by their racial and ethnic status, it is important to account for the socioeconomic conditions of neighborhoods: couples might migrate to avoid impoverished conditions, not simply because they prefer neighborhoods of certain racial and ethnic compositions.
Thus, in this study, I use longitudinal data collected across three decades from the Panel Study of Income Dynamics (PSID) linked to neighborhood- and metropolitan-level data from multiple population censuses to assess how disparate race and gender pairings of heterosexual black-white couples interact to shape their residential mobility and attainment of neighborhoods of varying ethnoracial composition. In particular, I investigate the propensity of black male–white female and white male–black female couples to move out of and into neighborhoods defined respectively by their levels of whites, blacks, and ethnoracial diversity, while controlling for micro-level and contextual characteristics. Investigating the residential mobility and attainment of these couples provides an opportunity to explore how traditional theories of residential stratification relate to a population that is a harbinger of what will be an increasingly diverse U.S. population in coming decades (Frey 2015).
Background and Theory
The study of both race and gender combinations of black-white couples is informed by a number of theoretical perspectives, the first of which comes from the life course perspective, which emphasizes the biographical, historical, and contextual variations in events and roles that span the length of life (Elder et al. 2015). Although the life course perspective consists of numerous principles, one of the more relevant for the study of the residential mobility and attainment of black-white couples with black or white male partners is the principle of linked lives, which highlights that lives are lived interdependently and that events and transitions are shared across a social network. In the context of this principle, researchers have observed that long-term romantic unions influence everything from individual self-rated health (Liu and Umberson 2011) to subjective well-being (Kamp Dush and Amato 2005). Given the evidence that coupling shapes individual life course experiences, it is reasonable to expect that the race and gender combination of individual members of couples, as well as their respective racial and gender status advantages and disadvantages, will combine to influence where couples live and, as a result, their exposure to neighborhoods of varying racial and ethnic compositions.
The linked lives of both race and gender combinations of black-white couples is further informed by research showing gender differences in residential mobility and attainment outcomes. The research on gendered preferences in residential mobility demonstrates that the likelihood of conducting a residential move is more strongly associated with the desires of male partners than with the desires of their female partners (Coulter et al. 2012). Other studies have observed that when heterosexual couples do move, women’s future employment status is negatively affected (Boyle et al. 2009; Cooke et al. 2009). These examples point to the possibility that gendered power disparities favoring men in residential decision-making could be extended to decisions related to the racial and ethnic composition of the neighborhoods to which and from which heterosexual couples migrate.
Gendered power disparities are particularly important to consider because preferences for neighborhoods of certain racial and ethnic compositions diverge by the racial and ethnic status of households (Farley et al. 1978), and this divergence contributes to broader patterns of racial and ethnic stratification (Bruch and Mare 2006). For example, when whites are asked to rank neighborhoods by attractiveness, they frequently select neighborhoods with the highest shares of whites and lowest shares of blacks as most attractive (Charles 2000). These findings are mirrored in research documenting that increasing shares of blacks are associated with a higher likelihood of whites migrating out of their neighborhoods (Crowder and South 2008). Not only are white households sensitive to increasing shares of blacks, but they also seem to avoid diverse neighborhoods when they move, selecting neighborhoods with high shares of whites (Crowder et al. 2012). Multiple scholars have asserted that whites’ avoidance of black and diverse neighborhoods is primarily based on prejudicial attitudes toward blacks and other ethnoracially diverse populations (e.g., Krysan et al. 2009); other research has contended that black and diverse neighborhoods serve as a proxy for negative neighborhood characteristics, such as high poverty, which tends to concentrate in these areas (Harris 1999). It is unclear, however, how the residential mobility of both race and gender combinations of black-white couples is influenced by the respective roles of neighborhood racial and ethnic composition and social class.
Scholars have also observed that blacks are more likely to prefer neighborhoods with more balanced levels of racial and ethnic diversity (e.g, Krysan and Farley 2002). Although diverse neighborhoods are increasing in number (Logan and Zhang 2011), they are still relatively uncommon. Thus, blacks frequently reside in neighborhoods with high concentrations of blacks. This residential mobility pattern could be driven partially by kinship geographies (Mulder 2007). Spring et al. (2017) found that kinship geographies play a significant role in the residential mobility and attainment patterns of households: having kin nearby lowered the likelihood of out-mobility, and those who did move were more likely to select a neighborhood destination that was nearby family than other neighborhood alternatives. In addition, households in poorer neighborhoods lived closer to parents than those residing in more-advantaged neighborhoods. These findings suggest that black-white couples with black male partners might be more apt to remain in and move to areas where his family resides, which would typically be in lower socioeconomic status black and diverse areas given the historical legacy of racial residential segregation (Massey and Tannen 2016).
Additionally, scholars have frequently asserted the central role of economic status in neighborhood attainment with the spatial assimilation theory, which argues that racial and ethnic minorities use their economic resources to gain entry into white—and often higher-income—neighborhoods, thus becoming spatially assimilated with the dominant majority group in the United States (Massey and Mullan 1984). There is some evidence that the economic resources of black-white couples are associated with the concentration of whites in their neighborhoods. For instance, Holloway et al. (2005) demonstrated descriptively that high-income black-white couples had higher shares of whites in their neighborhoods than low-income black-white couples. Gabriel (2016) observed that higher-income black-white couples had lower levels of racial and ethnic diversity in their neighborhoods than black-white couples with fewer resources. Similarly, Wright et al. (2013) found that higher-income black-white couples were more likely to reside in white neighborhoods and were less likely to be located in predominately black or diverse neighborhoods.
Another theoretical perspective is the place stratification theory, which states that racial and ethnic differences in neighborhood location are primarily driven by discrimination in the housing market (Massey and Denton 1993). Although declining, research has found that black households have encountered obstinate resistance to their presence in neighborhoods with large concentrations of whites (Sharkey 2013). The resistance to blacks migrating to predominantly white neighborhoods has been linked to the discriminatory actions of real estate agents (Turner et al. 2013), mortgage lenders (Rugh et al. 2015), and government policies (Rothstein 2017). Scholars have observed that even high-income black households are unable to convert their advantaged position into closer proximity with whites, leading them to assert that discrimination might be the primary cause of this stratification (Pais et al. 2012).
In investigating this, Logan and Alba (1993) established two variants of the place stratification theory. First, the strong version proposes that blacks obtain a weaker return on their economic resources compared with whites. Therefore, the most highly resourced blacks are posited to reside in areas with fewer whites than whites who possess relatively low levels of economic resources. Second, the weak version predicts that blacks are required to provide greater resources than whites to reside in whiter areas, meaning that the effect of economic resources is stronger for blacks than whites. Despite the stronger effect of income for blacks in this variant, they are still predicted to fall short of full neighborhood assimilation with the white majority.
Despite what is understood about discrimination toward blacks, how housing market discrimination affects black-white couples is not well known, and even less is known about discrimination toward black-white couples with black or white male partners. The little evidence available shows that to avoid the discrimination faced in nondiverse spaces, some black-white couples with black male partners seek black or diverse neighborhoods as places of refuge (Dalmage 2000:84, 103). Conversely, white men coupled with black women have been known to use their advantaged race and gender status to circumvent potential discrimination in the housing search and acquisition process (Romano 2003:131)—an option not available for black men who seek to obtain a place of residence for their households. Thus, the social advantages ascribed to white men could translate into black-white couples with white male partners possessing heightened abilities to reside in neighborhoods matching their preferences.
The aforementioned theories of residential stratification and prior studies on the topic of racial differences in residential outcomes provide useful insight into the potential patterns of residential mobility and attainment for black-white couples with black or white male partners. Based on this theoretical information and prior empirical analysis, I assert a general hypothesis that that the residential mobility and attainment patterns of couples with a white male partner will function similarly to that for white couples.
Data and Methods
To test this hypothesis, I use data from the Panel Study of Income Dynamics (PSID) linked contextual-level data from the U.S. Census. The PSID started in 1968 with a panel of approximately 4,800 U.S. families interviewed annually until 1997 and every two years thereafter. New families have been added to the panel as members of original panel families formed separate households. The PSID is advantageous for my analysis because its longitudinal structure makes it possible to track the residential location of individuals across time, and the data contain information on a variety of individual- and household-level characteristics known to influence residential mobility and attainment. Particularly useful, starting in 1985, the PSID began tracking the race and/or ethnicity of household heads and their spouses or long-term cohabitating partners.1 My sample thus includes the panel years between 1985 and 2015.
The PSID provides restricted-access Geospatial Match Files that allow me to link couples to their place of residence at each interview and attach comprehensive contextual information about their neighborhoods and metropolitan areas. I follow previous scholarship on the topic of residential mobility (Quillian 2002) by using census tracts as proxies for neighborhoods (Duncan and Duncan 1957), given that they are the most widely available spatial unit that approximates the typical conception of a neighborhood (Jargowsky 1997). Census data are drawn from the Neighborhood Change Database (NCDB; Geolytics 2013), in which census boundaries from 1980 to 2010 are normalized to 2010 census boundaries, providing consistent contextual measures over the time of the analysis. I use linear interpolation using NCDB census data from 1980, 1990, 2000, and 2010 to approximate neighborhood and metropolitan characteristics in noncensus years.
Each couple’s data record is segmented into a series of couple-period migration intervals, with each observation referring to the one- or two-year period between PSID interviews. Given my inclusion of metropolitan contextual characteristics, I further select couples that started and ended their migration interval in a census-defined metropolitan area. I also include only those householders with a spouse or a long-term cohabitating partner present at both the beginning and the end of an observation period.2 Given these data selections, my sample consists of 60,380 couple-period observations for mixed-race and monoracial couples. I include black-white couples consisting of two race and gender combinations: (1) couples composed of black males and white females (N = 604), and (2) couples consisting of white males and black females (N = 204). I also include two types of heterosexual monoracial couples: (1) both partners are white (N = 44,865), and (2) both partners are black (N = 14,707).
To better isolate couple category differences in residential mobility and attainment, I control for multiple theoretically relevant sociodemographic and socioeconomic characteristics. Characteristics include a predictor for age (in years) of the household head. I also include measures of household crowding, indicated by the number of persons per room; and length of residence, indicated with a dummy variable equal to 1 for those respondents who lived in their home for at least three years at the beginning of the observation period. The number of children under the age of 18 in the home is captured by a continuous predictor. I measure the financial resources of couples by their total family taxable income, measured in thousands of constant 2010 dollars. I control for education by measuring the household head’s number of completed school years. Additionally, I include measures for the employment status of the household head, coded as 1 for those employed at least part-time; and homeownership, coded as 1 for those residing in an owner-occupied housing unit.
Because the ethnoracial composition and socioeconomic status of neighborhoods are correlated (Massey and Tannen 2016), it is important to account for both characteristics in estimating couples’ patterns of residential mobility. For instance, those who migrate at higher levels of racial and ethnic diversity could actually be migrating to escape disadvantaged conditions associated with neighborhood poverty. Thus, when estimating residential mobility, I include a measure of the percentage of persons below the poverty line at the neighborhood level. Additionally, because neighborhoods with similar ethnoracial and socioeconomic compositions tend to cluster spatially (Massey and Denton 1993), which—in tandem with the distance-dependence of migration (Long 1988)—increases the likelihood of mobility into neighborhoods with similar ethnoracial and socioeconomic compositions, I control for the ethnoracial and socioeconomic composition of neighborhoods at the origin when measuring the racial and ethnic composition of destinations of mobile couples in my analysis. Additionally, each neighborhood-level variable is grand mean-centered to facilitate interpretation.
With these neighborhood-level predictors, I include measures of ethnoracial composition at the metropolitan level in estimating couples’ destination ethnoracial composition. Because couple category variation in residential attainment into neighborhoods defined by their racial and ethnic composition is partially determined by metropolitan opportunity structures, I include predictors of the metropolitan-level percentage of non-Hispanic whites, the metropolitan-level percentage of non-Hispanic blacks, and the metropolitan-level entropy score.
Characteristics included in the analysis are measured at the start of the observation interval and are considered time-varying. I include a measure for the year of observation to control for temporal trends in the ethnoracial composition of neighborhoods and mobility. I add a control variable for the length of the migration interval (one or two years) to account for the PSID adjustment to a biennial survey in 1997.
Last, although white-white couples serve as the reference category for the analysis, black-black couples are included in models to facilitate the understanding of the residential mobility and attainment patterns of black-white couples with black or white male partners to the group that has typically manifested the most distinctly disadvantaged residential outcomes compared with white-white couples. Hence, having both types of monoracial couples present in the analysis provides a frame of reference for the spectrum of residential mobility and attainment outcomes that both types of black-white couples might manifest.
Figure 1 presents the mean neighborhood percentage white, percentage black, and entropy for black-white couples with black or white male partners and monoracial black-black and white-white couples across all observation periods in my analysis. Figure 1 highlights that both types of black-white couples are typically exposed to more whites in their neighborhoods than are black-black couples, but these same black-white couples have fewer whites in their neighborhoods compared with white-white couples. Black-white couples with black male partners have roughly similar levels of whites in their neighborhoods (56.89 %) compared with their counterparts with white male partners (55.99 %). Consistent with past studies (Logan and Stults 2011), white-white couples exhibit extremely low percentages of blacks in their neighborhoods (5.72 %), while the average black-black couple has roughly 10 times more blacks in their neighborhoods (55.07 %). Again, black-white couples with black or white male partners emerge with similar levels of exposure to blacks in their neighborhoods, at 29.14 % and 30.37 %, respectively. In terms of neighborhood entropy, black-white couples with black male partners (51.80) and those with white male partners (53.80) possess the highest levels of diversity in their neighborhoods, followed by black-black couples, with white-white couples demonstrating the lowest exposure to diversity.
Table 1 provides the descriptive statistics disaggregated by couple categories for the variables used in the analysis. Of all couples, black-white couples with black male partners possess the highest likelihood of changing tracts between interviews (25.99 %), followed by black-white couples with white male partners (20.10 %) and black-black couples (18.18 %); white-white couples demonstrate the lowest likelihood of residential mobility (14.17 %). As expected, white-white couples reside in metropolitan areas with the greatest shares of whites, followed by black-white couples with black male partners and then those with white male partners; black-black couples live in areas with the lowest white shares. This pattern is inversed for the percentage of blacks at the metropolitan-level: black-black couples possess the highest concentrations of blacks, where both types of black-white couples have similar levels, and whites have the lowest concentrations. Given these previous statistics, it is not surprising that couples with black partners live in metropolitan areas with the greatest diversity.
Table 1 also highlights sharp couple category differences in micro-level characteristics. Compared with white-white couples, both types of black-white couples are younger, tend to have more children, live in more crowded dwellings, and are less likely to reside in the same house for at least three years. Pronounced differences in family income also emerge between couples with a white male partner and those without. For instance, black-white couples with white male partners earn approximately $83,790 per year; white-white couples earn approximately $80,300 annually; black-white couples with black male partners earn $53,240 annually; and black-black couples earn a yearly family income of $51,510. Both types of black-white couples are less likely to be homeowners compared with both white-white and black-black couples, are more apt to be employed than black-black couples, and possess relatively similar levels of education as white-white couples.
Patterns of Residential Out-Mobility
Even though group differences in neighborhood racial and ethnic composition between black-white couples with black or white male partners are not substantively different, each type of black-white couple may follow variant residential out-mobility and destination patterns, leading them to neighborhoods with similar racial and ethnic compositions. Hence, the focus in these models is the differential effect of neighborhood percentage white, percentage black, and entropy, respectively, on the likelihood of out-mobility across various couples.6 To investigate these patterns, I use three-level random-intercepts logistic regression models predicting the log odds of leaving the neighborhood of origin between sequential PSID interviews.7
In Model 1 of Table 2, I use product terms that allow me to examine the influence of larger white shares in the neighborhood of origin on residential out-mobility for all couples in my analysis. With the inclusion of the control variables, the coefficient for neighborhood percentage white—representing the effect of higher white shares on white-white out-mobility—is negative and significant (b = –0.0083, p < .001). Black-black couples, however, emerge with a positive and significant coefficient (b = –0.0083 + 0.0119 = 0.0036, p < .001). As it relates to black-white couples, the coefficient for those with white male partners fails to reach statistical significance. Conversely, the effect of higher white shares in the neighborhoods of black-white couples with black male partners is statistically significant; thus, a 1 standard deviation increase in the share of whites is associated with a 28 % [e((–0.0083+0.0168) × 29.492) = 1.2846] increase in the odds of out-mobility compared with white-white couples.
To gain further insight into the product terms presented in Model 1, panel a of Fig. 2 presents the predicted probability of out-mobility by each type of couple in my analysis by the percentage white at the neighborhood-level (restricted to values within the 10th and 90th percentiles), while the remaining covariates are held at their means.8 Panel a of Fig. 2 highlights that the probability of out-mobility for white-white couples is low no matter the concentration of whites in their neighborhoods. In contrast, black-black couples display a positive probability of out-mobility as their neighborhoods contain higher concentrations of whites. Congruent with theoretical expectation, black-white couples with white male partners emerge with a similar negative slope as white-white couples; however, the probability of out-mobility for black-white couples with white male partners is higher across all points of their slope. In the case of black-white couples with black male partners, they are likely to migrate out of neighborhoods containing higher concentrations of whites, possibly in an attempt to avoid discrimination.
Model 2 of Table 2 adds product terms allowing an analysis of the association between the percentage black in the origin neighborhood of couples on their likelihood of changing tracts between interviews, net of controls. White-white couples with larger shares of blacks in their neighborhoods demonstrate a positive and significant coefficient for out-mobility (b = 0.0078, p < .001). Conversely, black-black couples are less likely than white-white couples to move when their neighborhoods consist of large shares of blacks (b = 0.0078 – 0.0119 = –0.0041, p < .001). The coefficient for black-white couples with white male partners is not statistically significant. Black-white couples with black male partners evince a statistically significant and negative association between larger black concentrations within their neighborhoods and residential out-mobility. For these couples, a 1 standard deviation increase in the percentage of blacks within their neighborhoods is related to a 25 % [e((0.0078–0.0182) × 28.043) = .7470] decrease in the odds of out-mobility.
Panel b of Fig. 2 displays the predicted probability of out-mobility by each type of couple and percentage black at the neighborhood-level (restricted to values within the 10th and 90th percentiles), while holding the remaining covariates at their mean values. White-white couples demonstrate a positive slope of out-mobility with higher concentrations of blacks in their neighborhoods. In the case of black-black couples, their negative slope highlights their decreasing residential mobility in neighborhoods with higher shares of blacks. Black-white couples with white male partners emerge with a higher probability of out-mobility at all points of their distribution than white-white couples but with a similar positive slope. This finding aligns with the theoretical expectation that the preferences of white men in relationships with black women for neighborhoods with lower shares of blacks are relatively similar to white-white couples. Black-white couples with black male partners, however, display the steepest negative slope, with larger concentrations of blacks in their neighborhoods. This finding provides possible support for the notion that black-white couples with black partners remain in predominantly black neighborhoods because their inhabitants are more accepting of their unions.
Model 3 of Table 2 includes control variables and product terms for each type of couple and neighborhood entropy. White-white couples emerge notably with the highest likelihood of out-mobility with larger concentrations of neighborhood diversity (b = 0.0082, p < .001). Black-black couples, however, reveal a lower likelihood of out-mobility in relation to higher neighborhood diversity (b = 0.0082 – 0.0063 = 0.0019, p < .001). For both types of black-white couples, their interactions with neighborhood diversity are statistically nonsignificant.
Panel c of Fig. 2 illustrates the predicted probability of out-mobility by each type of couple and neighborhood diversity (restricted to values within the 10th and 90th percentiles), while holding the remaining covariates at their mean values. All couples in the analysis demonstrate an increasing probability of out-mobility associated with higher levels of diversity in their neighborhoods. White-white couples have the lowest intercept of out-mobility of all couples, and they also reveal the shortest range of entropy values given that they typically reside in neighborhoods with the lowest diversity. Black-black couples display a slightly positive slope for out-mobility across their entire distribution. Black-white couples with white male partners emerge with the strongest positive slope out of all couples, which suggests that the probability that these couples will change tracts between interviews is higher in neighborhoods with substantial diversity. Black-white couples with black male partners display the highest probability of out-mobility of all couples in the analysis, a finding counter to theoretical expectation but reflective of their higher mean level of out-mobility between interviews (see Table 1).
Neighborhood Racial and Ethnic Composition for Mobile Couples
Theories of residential mobility and attainment not only highlight the importance of residential out-mobility in explaining group differences in residential location, but they also point to the selection of destinations as central in understanding broader patterns of racial stratification. Therefore, these models assess the percentages of whites, blacks, and the level of diversity, respectively, in the destination neighborhoods of mobile couples. Additionally, I test the effect of economic resources on mobility into the aforementioned neighborhoods. I use three-level random-intercepts linear regression analysis and account for the nonrandom selection of individuals into the mover category through a Heckman correction strategy (Heckman 1979).
In Model 1 of Table 3, I observe significant couple differences in the percentage white in their destination neighborhoods, net of controls. For instance, black-black couples have the lowest shares of whites in their destination neighborhoods (b = –25.3057, p < .001). Of all couples with a black partner, black-white couples with white male partners exhibit the highest concentrations of whites in their destination tracts (b = –14.5232, p < .001). Black-white couples with black male partners (b = –16.1381, p < .001) reside in neighborhoods with significantly lower concentrations of whites compared with white-white couples.
Model 2 of Table 3 tests the spatial assimilation theory that economic resources assist couples in migrating to neighborhoods with higher concentrations of whites, while controlling for associated covariates. The effect of economic resources—as measured by family income—on increasing shares of whites in the destination neighborhoods of white-white couples is positive and significant (b = 0.0086, p < .05). Black-black couples also evince a positive and significant effect (b = 0.0086 + 0.0873 = 0.0959, p < .001) of family income on the shares of whites in their destination neighborhoods. The effect of family income on the percentage of whites in the destination neighborhoods of black-white couples with white male partners fails to reach statistical significance. A similar nonsignificant effect is witnessed for black-white couples with black male partners.
To further clarify the role of economic resources on the concentration of whites in the neighborhoods where couples migrate, panel a of Fig. 3 illustrates the predicted percentage of neighborhood white by couples and family income (restricted to values within the 10th and 90th percentiles), while the remaining covariates are held at their mean values. Immediately apparent is the presence of high concentrations of whites in the destination neighborhoods of white-white couples, no matter their level of family income. In line with the weak version of the place stratification theory, black-black couples demonstrate a stronger positive association between their level of family income and the amount of whites in the neighborhoods where they migrate. Black-white couples with white male partners and those with black male partners also follow the weak version of the place stratification theory. Both types of couples demonstrate a similar positive association between family income and the concentrations of whites in the neighborhoods where they migrate, seemingly occupying a middle space between black-black and white-white couples.
Model 3 of Table 3 assesses the percentage black in the destination neighborhoods of the couples in the analysis, while controlling for relevant covariates. Black-black couples exhibit the highest concentration of blacks in their destination neighborhoods (b = 22.9587, p < .001) relative to white-white couples. Black-white couples with black-male partners show higher percentages of blacks in their destination neighborhoods (b = 12.5590, p < .001) than white-white couples. Black-white couples with white male partners have the lowest exposure to blacks of all couples with a black partner (b = 10.1116, p < .001) but experience larger concentrations than white-white couples.
Model 4 of Table 3 investigates how the effect of family income varies for couples in their attainment of neighborhoods defined by their concentrations of blacks, net of controls. The effect of family income on the percentage of blacks that white-white couples are exposed to in their destination neighborhoods is null. Thus, regardless of their income level, white-white couples have minimal black shares in the neighborhoods where they migrate. Given that black households state preferences for neighborhoods with more balanced representations of black and nonblack racial and ethnic groups (Charles 2003), the finding that enhanced levels of family income are associated with fewer shares of blacks in their destination neighborhoods is not surprising (b = –0.0016 – 0.1066 = –0.1082, p < .001). In regard to black-white couples with white male partners, the effect of family income on the percentage of blacks in their destination neighborhoods fails to reach statistical significance; the same is true for black-white couples with black male partners.
Panel b of Fig. 3 displays the predicted percentage of neighborhood black by couples and family income (restricted to values within the 10th and 90th percentiles), while holding the remaining covariates at their means. Confirming the findings of Model 4 of Table 3, panel b of the figure illustrates that white-white couples, even those with low levels of family income, have markedly small black representations in their destination neighborhoods. Additionally, only the highest-earning black-black couples are associated with lower concentrations of members of their race at the neighborhood-level. Similar to black-black couples, black-white couples with black male partners tend to purchase their way into neighborhoods with lower shares of blacks. Their counterparts with white male partners demonstrate a slight positive slope but still emerge with a relatively stable percentage of blacks in their destination neighborhoods at all points along their income distribution.
Model 5 of Table 3 investigates couple differences in the level of neighborhood diversity for mobile couples, while controlling for applicable covariates. Black-black couples demonstrate a positive and significant (b = 0.0866, p < .001) association with diversity in their destination neighborhoods compared with white-white couples, potentially signaling their preferences for meaningful levels of diversity. Couple differences are most pronounced for black-white couples with black male partners, emerging with a positive and highly significant (b = 0.1058, p < .001) association with neighborhood diversity relative to white-white couples. The coefficient for black-white couples with white male partners is positive and significant (b = 0.0775, p < .05), but they migrate to the lowest-diversity neighborhoods of all couples with a black partner.
In Model 6, I include product terms to evaluate the degree to which the effect of economic resources for couples is associated with the level of diversity in their destination neighborhoods, net of controls. For white-white couples, the effect of family income on destination neighborhood diversity is negative and significant (b = –0.0001, p < .01), a finding consistent with past research showing that whites attempt to avoid nonwhite populations when migrating (Crowder et al. 2012). Standing in contrast, and in line with stated preferences for diversity from past research (Charles 2000), black-black couples tend to use economic resources to purchase their way into neighborhoods with higher diversity (b = –0.0001 + 0.0008 = 0.0007, p < .001). Moreover, the effect of family income on neighborhood diversity for black-white couples with white male partners is statistically nonsignificant. A similar effect of family income emerges for black-white couples with black male partners, for whom the product term is nonsignificant.
Panel c of Fig. 3 highlights the predicted level of neighborhood diversity by couples and family income (restricted to values within the 10th and 90th percentiles), while holding the remaining covariates at their means. Similar to panel b, high-income white-white couples are associated with neighborhoods possessing relatively low levels of diversity. Also clear is the general clustering of intercepts and slopes of those couples with a black partner. Among this clustering, however, the slopes deviate. Black-black couples show a positive slope, again potentially signaling a preference for neighborhood diversity (Clark 2007). Both types of black-white couples exhibit a negative association between increased economic resources and neighborhood diversity. Yet, given their high intercept values and weakly negative slopes, these couples still display high levels of diversity across their income distributions compared with white-white couples, a finding congruent with previous research observing that these couples typically reside in areas of high diversity (Wright et al. 2013).
Researchers have pointed to the need for a broader understanding of how race and gender are associated with the residential patterns of black-white couples (e.g., Wright et al. 2013). My analysis builds on this prior research by using 30 years of longitudinal data from the PSID linked to neighborhood- and metropolitan-level data to assess black male–white female and white male–black female couples’ mobility out of and into neighborhoods defined by their percentages of whites and blacks and their ethnoracial diversity.
The results indicate that the race of the male partner aligns with the ethnoracial composition of the neighborhoods where both types of black-white couples remain. Black-white couples with white male partners are likely to remain in predominantly white neighborhoods but tend to exit neighborhoods that are black or ethnoracially diverse. However, their tendency to remain spatially assimilated in predominantly white neighborhoods means that black female partners are engaging in frequent cross-racial contact, potentially lessening prejudicial attitudes among whites (Allport 1954). Conversely, black-white couples with black male partners are likely to move out of neighborhoods with high white shares, but they tend to remain in black areas. These residential mobility patterns potentially represent the discomfort that these couples experience in predominantly white neighborhoods due to discrimination, as posited by the place stratification theory. Unexpectedly, black-white couples with black male partners are likely to leave highly diverse areas, perhaps because of the lower average ages of these couples increasing the likelihood of their mobility in general. Future research should investigate age-related neighborhood migration patterns among these couples to provide insight into whether their exposure to neighborhood diversity is age-dependent.
The general pattern of migration among both types of black-white couples points to higher levels of racial and ethnic diversity in their destinations. However, the effect of economic resources on the racial and ethnic composition of their destination neighborhoods appears relatively modest, potentially pointing to other explanations beyond the role of socioeconomic status in shaping their residential attainment. For instance, kinship geographies might be salient in explaining these outcomes. The relative distribution of kin across neighborhoods is likely shaped by a legacy of racial residential segregation; thus, black respondents will have kin in areas that are more black and diverse, while whites are more likely to have kin in neighborhoods of higher concentrations of whites. Consequently, when both types of black-white couples migrate, they might search for more-diverse spaces that buffer between predominantly black and predominantly white areas (Logan and Zhang 2010) where their respective kin reside. However, as stated earlier, some households live in poor areas to be close to kin (Spring et al. 2017), which could be influential for policy efforts to further integrate neighborhoods because there are likely black-white couples who would prefer to live in diverse neighborhoods but remain in, or move to, poorer black neighborhoods to aid low-income kin. Future policy efforts to promote neighborhood integration must grapple with the fact that household residential decision-making is embedded in an interdependent social network, as stressed by the life course principle of linked lives.
The findings from this study point to the need for scholars to be mindful of diverse household racial and ethnic compositions. Given the increasing levels of household racial and ethnic diversity in the United States (Wang 2012), traditional theories of residential stratification need to adapt to this new reality. The new reality of increased household racial and ethnic mixing brings gender into relief, necessitating that theories originally designed around the central role of race in determining residential location expand to include gendered power disparities. The intersectional relationship between race and gender in residential stratification is also conditioned by historical legacies that ascribe advantages and disadvantages to couples by race and gender, as suggested by the life course principle of linked lives. Thus, the life course of individuals in couples is likely influenced by their respective partners’ unique racial histories that shaped their partners’ preferences for certain types of neighborhoods. These racial preferences for certain neighborhoods combine with the gender status of individuals in couples to partially condition both how they navigate the housing market and how actors and institutions in the housing market relate to diverse couples.
Understanding the unique residential patterns of black-white couples with black or white male members would be further enhanced by systematically measuring their residential preferences for neighborhoods of specific racial and ethnic compositions, while also tracking their residential mobility into and out of neighborhoods across their life course. Future research should also investigate the residential mobility and attainment of these couples into neighborhoods that follow distinct trajectories of neighborhood racial and ethnic change: this might inform our understanding of these couples’ preferences for changing neighborhoods. Moreover, scholars would benefit by using discrete choice models. This approach allows for the consideration of how multiple dimensions of neighborhoods, such as socioeconomic status and racial composition, simultaneously influence couples’ selection of an eventual destination (Quillian 2015). In total, these ideas would help to expand future studies on the various race and gender combinations of mixed-race couples and enhance our understanding of how increasing numbers of diverse households function within long-established systems of residential stratification.
I thank Kyle Crowder, Stewart Tolnay, Charles Hirschman, Mark Ellis, Christina Hughes, Tim Heaton, Cardell Jacobsen, the Editors of Demography, and anonymous reviewers for their very helpful comments on earlier versions of this article.
The PSID defines long-term cohabiters as those coupled with a sample member with whom they have shared a residence for at least 12 months.
Members of couples whose relationships end during the study period are removed from the analysis for the remaining years. However, they may return to the sample if they form another union.
The other category consists of Asians, Native Hawaiians and Pacific Islanders, American Indians and Alaska Natives, those who claim multiracial status, and some other race.
Models are estimated using the xt suite commands in Stata 14 (StataCorp 2015).
Random coefficients are not estimated in both the three-level random-intercepts logistic and the three-level random-intercepts linear regression models in this analysis.
With racially mixed people appearing in the U.S. Census beginning in 2000, measures of entropy could be affected. I conducted a supplemental analysis in which I excluded those who claimed a mixed-race status in the 2000 and 2010 censuses from the entropy measures and then reestimated the models presented in Tables 2 and 3. The results from this supplemental analysis are highly similar to those reported in the article and are available in Online Resource 1.
The use of census tracts as proxies for neighborhoods raises questions about the Modifiable Areal Unit Problem in the likelihood of residential out-mobility across couple categories given that moves of a given distance are more likely to be classified as intertract if the tracts are small. Supplemental analysis reveals that controlling for tract size in square miles in the models predicting the likelihood of out-mobility does not alter the substantive findings of the results presented in Table 2. See Online Resource 1 for these results.
In an analysis available in Online Resource 1, I explored whether the predicted values presented in Figs. 2 and 3 differ if the remaining covariates are held at their means or their observed values. The predicted values for the out-mobility models across couple categories with covariates held at their means are all lower than when the covariates are at their observed values. The predicted values for ethnoracial composition in destination neighborhood when family income is allowed to vary are highly similar when the covariates are held at their means or estimated using observed values.