Abstract
We use data from a random sample of students collected at two large public universities, one in the Midwestern region and one in the Southeastern region of the U.S., to document the prevalence of self-reported “interracial” romantic relationships (SR-IRRs) and the extent to which self-identifications differ from researcher-defined categorizations that label as “interracial/interethnic” all partnerships that cross racial or Latino/Hispanic ethnic boundaries (RD-IRRs). Our findings show that a substantial percentage of students in relationships that cross racial/ethnic lines do not identify them as “interracial.” As a result, measures of SR-IRR and RD-IRR produce very different prevalence figures for cross-group relationships (SR-IRR = 18% of respondents; RDIRR = 24%). The disjuncture between self-reports and researcher-defined categorizations is particularly pronounced for Hispanics and, to a lesser degree, non-Hispanic Whites. The consistency with which relationships that include non-Hispanic Black individuals are labeled, however, stands out as unique: every instance of racial/ethnic boundary crossing that involved an individual from this group was labeled as “interracial.” Multivariate analyses identify significant predictors of SR-IRRs and RDIRRs and how race/ethnicity, nativity status, and university location interact to shape relationship engagement and the likelihood of self-identifying relationships as “interracial.” Our discussion concludes with implications and suggestions for future research.
Keywords
More people in the United States than ever are crossing racial and ethnic lines to engage in romantic interracial or interethnic relationships (IR/IERs). Between 1980 and 2015, intermarriages grew from 3 to 10% of all marriages, and by 2015 at least 17% of newly married couples were interracial/interethnic (IR/IE) (Livingston and Brown 2017). Reliable statistics on non-marital relationships are elusive, but research suggests that IR/IE cohabitation and dating are even more common than IR/IE marriage, and that the increase in these relationships may be particularly pronounced among youth and young adults (Levin, Taylor, and Caudle 2007; Miller 2014; Tillman and Miller 2017). Nearly 20% of adolescents now report that they have crossed racial/ethnic boundaries for romance (Kreager 2008; Miller 2014; Tillman and Miller 2017; Wang et al. 2006) and between 24% and 60% of college students have dated at least one member of another group (Keels and Harris 2014; Levin et al. 2007). Young adults also hold more positive attitudes than older adults about IR/IE relationships, as those ages 18–29 are the least likely to oppose interracial marriage and most likely to believe that interracial marriage is “good for our society” (Livingston and Brown 2017).
Accurately categorizing IR/IE partnerships has been challenging however, as past researchers have used different methodologies for identifying such couples. Most recent studies create a researcher-defined measure of IR/IERs that includes those who cross either traditional racial lines or Latino/Hispanic ethnic lines (or both) by comparing respondents’ self-reports of their own and their partners’ race and ethnicity. These methodological decisions reflect the general trend over time by which scholars (Harris and Ono 2005) and the general public (Parker, Horowitz, Morin and Lopez 2015) have increasingly treated a Latino/Hispanic background (henceforth referred to simply as Hispanic) as “racial” in nature and viewed those who are Hispanic as belonging to a group distinct from any traditional racial group.
Yet, research suggests that the acceptance of a racialized understanding of Hispanic ethnicity is not universal in the U.S. (Perlmann and Waters 2002), particularly among those who identify as Hispanic. Many in this group categorize themselves as “some other race” and consider their Hispanic background as part of both their race and ethnicity (Parker et al. 2015). Research also suggests that social distance between those of Hispanic background and individuals of other groups varies, with non-Hispanic Whites often viewed as less “different” from Hispanics than non-Hispanic Blacks or Asians (Michalikova and Yang 2011). However, we do not have a clear understanding of how Hispanic ethnicity factors into the way that young people understand and label “racial” differences between themselves and their partners and we can find almost no recent studies that ask respondents directly whether they self-identify as being in an “interracial” relationship (Perry 2013, 2014).
As such, researcher-defined measures of IR/IERs may not reflect young adults’ own understandings of whether they are engaging in a boundary-crossing relationship. The use of researcher-defined measures may lead scholars to generate statistics that indicate a higher prevalence of cross-group relationships than would the use of self-reported measures. Moreover, the use of researcher-defined measures that combine into one category respondents who consider themselves to be in an interracial relationship with those who do not may muddy our understanding of how engagement in an IR/IER is related to other social and emotional outcomes. The extent to which findings using self-reported measures and researcher-defined measures diverge, however, is an open question.
The present study uses survey data collected through random sampling at two large predominantly White public universities to address these issues. We begin by documenting the prevalence of self-reported “interracial” romantic relationships (SR-IRRs) among college students. We then explore the extent to which self-identifications differ from researcher-defined categorizations that label as “IR/IE” all partnerships that cross either racial or Hispanic ethnic boundaries (RD-IRRs). Multivariate analyses examine the factors associated with SR-IRRs and RD-IRRs, as well as the extent to which race/ethnicity, university location, and nativity status interact to shape individuals’ engagement in relationships across racial/ethnic lines and their likelihood of self-identifying that engagement as “interracial.” Finally, we discuss the implications of our findings for our understanding of the prevalence and potential meaning of interracial relationships among this age group, as well as suggestions for future data collection efforts and avenues of research.
Background
IR/IE Romance Among College Students
Contemporary college students represent a unique group for examining IR/IERs, as this generation of youth is much more diverse than previous generations (Fry and Parker 2018), and, for many individuals, the late teens and early twenties are the years during which romantic involvements increase precipitously (Olmstead and Anders 2021). The diversity found on many college campuses allows students to interact with people from backgrounds different than their own and sometimes, particularly for non-Hispanic White students, their first close exposure to members of different racial/ethnic groups (Bowman and Park 2014). College campuses also often provide social environments that allow students freedom from the rules and expectations of parents and other family members. Although most college students believe their parents generally approve of interracial dating, most recent college students do not have first-hand examples of IR/IE relationship behavior within their families of origin (Miller et al. 2022; Roebuck and Murty 2015). As such, college provides a relatively free environment within which to develop new relationships, and it is not uncommon for college students to form cross-group romantic partnerships. Indeed, between 24% and 60% of college students have dated at least one member of another group (Field, Kimuna, and Straus 2013; Keels and Harris 2014; Levin et al. 2007; Miller et al. 2022).
There are significant racial and ethnic differences, however, in the prevalence of cross-group partnerships and the particular patterns of racial composition seen among those in IR/IERs. For example, young people who identify as Asian and Hispanic have an especially high likelihood of engaging in cross-group relationships (Keels and Harris 2014; Levin et al. 2007), whereas Non-Hispanic Black and Non-Hispanic White youth are much less likely to do so (Lee and Bean 2010; Livingston and Brown 2017). Among those within IR/IERs, people of color are most likely to report having a White partner (Keels and Harris 2014; Roebuck and Murty 2015) and White individuals are most likely to form relationships with Hispanic or Asian partners (Keels and Harris 2014).
Geographic region and opportunities for intergroup contact, particularly between White, Hispanic, and Asian individuals, also appear to be strong predictors of IR/IER attitudes and behavior. For example, adults living in the South and Midwest regions of the U.S. tend to be less accepting of IR/IERs (Livingston and Brown 2017). Relatedly, intermarriage is least prevalent in parts of the South and in the Midwest and most prevalent in Western urban areas, particularly those that are more racially/ethnically diverse (Livingston and Brown, 2017).
The Difficulties of Conceptualizing and Measuring IR/IE Relationships
Despite the growing body of literature that has focused on IR/IERs, measuring engagement in these relationships has remained challenging. Studies generally do not ask respondents whether they view their relationships as having crossed salient racial or ethnic lines. Rather, most recent studies measure IR/IERs by comparing respondents’ reports of their own race and Hispanic ethnicity and their partners’ race and ethnicity to create a measure that includes those who cross either traditional racial lines or Hispanic ethnic lines, despite the fact that a racialized understanding of Hispanic ethnicity is not universal (Parker et al. 2015; Rumbaut 2011). These measures also generally ignore other potential ethnic differences between partners (e.g., ethnic variations within the Asian American populations), which some respondents might view as salient to the “racial” categorization of their relationship. These methodological decisions, combined with the fact that racial and ethnic identities can be fluid and hold different meanings for different groups of people (Gonzalez-Barrera and Lopez 2015), mean that researcher-defined measures may not reflect young adults’ own understandings of whether they are engaging in a boundary-crossing relationship.
Racial/ethnic identity development is a fundamental part of many students’ experiences during their college years (Long 2012; Patton 2016), particularly for youth of color. While various models have been advanced to understand the development of racial/ethnic identity, scholars generally agree that it occurs as a process whereby young people, as a result of increasing experience with those from other groups, move from lack of awareness and/or rejection of their differences with others to a stronger acceptance of and attachment to a collective group identity (Phinney, Chavira and Tate 1993; Sue and Sue 2003). A strong group identity necessarily highlights the boundaries of group membership and may be an important factor in whether or not someone labels a relationship as having crossed a salient group boundary.
In the U.S., there is a clear social consensus that the categories of “Asian,” “Black,” “Native American,” and “White” are distinct “racial” groups. The increasing number of young people who identify as multiracial (U.S. Census 2021; Parker et al. 2015) complicates and adds nuance to this simplified categorization process and may lead to more complexity and variation in the process of identity development and in the way that people conceive of “racial” boundary-crossing (Bratter 2007; Buggs 2019; Littlejohn 2019). Moreover, in the U.S., the Hispanic category is treated variously as an ethnic designation that is distinct from one’s racial background (e.g., one might identify as Hispanic White and Hispanic Black) and as a racialized category unto itself (e.g., one might identify as Hispanic or non-Hispanic White) (Perlmann and Waters 2002; Parker et al. 2015).
This lack of consensus leads to greater variation in how Hispanic individuals identify themselves racially (Perlmann and Waters 2002; Rumbaut 2011) and, as result, may influence whether they (and their partners) view their Hispanic background as being salient to the categorization of their relationship. Many young adults in Hispanic/Non-Hispanic partnerships may consider themselves to be in a same-race relationship, while others may consider the same type of relationship to be “interracial.” Yet, it remains unclear how the majority of college students would categorize a Hispanic/Non-Hispanic relationship and whether this differs by university location.
Understanding Differential Engagement in and Recognition of IR/IERs
Marginalization and social distance
Despite the rising prevalence of IR/IERs, support for interracial relationships is not universal. Research focused specifically on college students has found that large percentages of both White and Black students still are not open to the idea of personal involvement in an IR/IER (Joyner and Kao 2005; Knox et al. 2000). Given the persistence of negative and/or resistant attitudes toward IR/IERs, being in one can be viewed as occupying a marginalized social position. The extent of the marginalization experienced as a result of this social position may vary, however, by the particular racial/ethnic combination of the partners.
Some minority groups experience more social distance from the majority group than others (Michalikova and Yang 2011), making intergroup relationships that include them more socially taboo. Individuals who engage in these relationships are likely to face a more marginalized social position and more exposure to social stressors such as racism, discrimination, and stigmatization (Bonilla-Silva and Forman 2000; Datzman and Gardner 2000; Solsberry 1994), all of which can have detrimental effects on a person’s mental health and relationship outcomes (Miller, Rote, and Keith 2013). Thus, in addition to issues of identity development, young people’s participation in and self-reports of engagement in “interracial” relationships may be shaped by the differential levels of social acceptance associated with particular cross-group combinations (Herman and Campbell 2012; Passel, Wang, and Taylor 2010; Perry 2014).
In the U.S. context, non-Hispanic Whites and Hispanics often are viewed as less “different” from each other than they are from non-Hispanic Blacks or Asians (Michalikova and Yang 2011; Muro and Martinez 2016). This attitude may reflect the widespread perception that these two groups are culturally and phenotypically more similar to one another and tend to overlook the existence of a growing Afro-Latino population (Feliciano and Robnett 2014; Haywood 2017; Lopez and Gonzalez-Barrera 2016; Morales 2012). It may also encourage the formation of relationships between these two groups, as well as greater variability in whether they are viewed as “interracial.” Conversely, Lee and Bean (2010, 2012) have highlighted an enduring pattern of “Black exceptionalism.” While research documents a rapid rise in Black–White intermarriages, Black adults and adolescents still are much less likely than Asians or Hispanics to participate in interracial unions, particularly with Whites (Kreager 2008; Lee and Bean 2010; Qian and Lichter 2007; Wang et al. 2006). Studies show that Whites, Asians, and Hispanics tend to hold similar racial exclusion preferences, often excluding Black individuals as potential romantic partners (Feliciano and Robnett 2014; Robnett and Feliciano, 2011). In addition, among college students of all racial/ethnic groups, approval of Asian/White relationships is higher than approval of Black/White relationships (Field et al. 2013). Less is known about college students’ attitudes toward relationships that cross Hispanic ethnic lines. In general, though, the literature suggests that participating in a cross-group relationship that includes a Black partner may be more taboo and associated with greater social stressors than other cross-group combinations (Baptist, Craig, and Nicholson 2018; Livingston and Brown 2017; White et al. 2015). Recognition of their marginalized treatment may increase the salience of this particular racial boundary crossing, thereby reducing the likelihood college students will engage in an IR/IER that includes one Black partner and increasing the likelihood that, if they do so, they will consistently label it as “interracial.”
Nativity status may also shape understandings of “race,” social distance, and boundary crossing. High proportions of Hispanic and Asian young adults are immigrants to the U.S. (Patten 2016; U.S. Census 2014). These young people hail from countries in which notions of “race” differ significantly from notions in this country (Dikötter 1997; Telles 2014: Wade 2017), leading to potentially more variation in how they view their own race and ethnicity and how they categorize their romantic relationships (Wade 2010). Classical assimilation theories suggest that social distance between immigrants and non-immigrants tends to decline over time and across generations (Gordon 1964; Wu et al. 2015), leading to more relationships with native-born partners for young immigrants whose families have been in the US for a longer period of time (Levin et al. 2007). At the same time, greater assimilation may lead to a stronger adoption of U.S. norms in regards to racial/ethnic categorization and interracial/ethnic dating behaviors, reducing the likelihood of relationships that cross racial/ethnic lines and increasing the consistency with which boundary-crossing relationships are labeled as “interracial.” Reformulations of assimilation theory (e.g., Alba 2020) have also recognized our society’s increasing embrace of multiculturalism and the uneven incorporation of various groups into the American social order, both of which may mean that the effect of nativity is conditioned by other background and contextual factors, like race/ethnicity and location.
Social contact theory and location
The processes of identity development and intergroup assimilation are also influenced by the type and level of social contact experienced between racial/ethnic groups, and this may vary substantially across regions and specific geographic locations (Pettigrew et al. 2011). In particular, research suggests that intergroup contact can lead to either increased harmony or conflict between groups, depending upon the context in which the social interaction takes place (Muro and Martinez 2016). Social Contact Theory proposes that racial prejudice generally is reduced in situations where there is long-term and meaningful social contact between members of different groups that have equal status, share common goals, perceive institutional support for racial mixing, and have the opportunity to develop meaningful cross-group friendships (Allport 1954; Clayton 2019; Pettigrew et al. 2011). Ideally, university campuses that include students from a variety of racial and ethnic backgrounds provide such conditions. The reality, however, is that present-day campuses are not color-blind spaces in which all groups have equal status (Laybourn et al. 2017). Universities are embedded in broader geographic areas and specific city/town locations that have unique racial climates and histories. Moreover, students bring with them to campus their own previous racialized experiences, identities, and attitudes. Together, these forces affect the potential for individuals to develop interracial friendships and romantic relationships (Laybourn et al. 2017; Pettigrew 1998).
Indeed, cross-group interaction alone may not be sufficient for breaking down the social distance between students from different racial/ethnic groups. In particular, predominantly White institutions (PWIs) may struggle to provide the optimal conditions for meaningful intergroup contact between Black and non-Black students (Clayton 2019), even when other racial/ethnic groups are engaging in meaningful contact. Variation almost certainly also exists between PWIs with different histories and different racial/ethnic compositions.
Blau’s macrostructural theory of inter-group relations stresses the importance of organizational composition, positing that a group’s relative size is inversely related to its propensity to form outgroup relationships (Blau, Blum, and Schwarz 1982). Research on racial/ethnic intermarriage generally supports this theory (Hou, Wu, Schimmele, and Myles 2015). These dynamics may also hold for the formation of IR/IERs on university campuses. Within universities, students from relatively smaller racial/ethnic groups may be more likely to form out-group relationships than those from larger racial/ethnic groups, and, when comparing two different universities, prevalence rates for a particular racial/ethnic group may be higher at the university in which they make up a relatively smaller portion of the student population.
Study Aims
The purpose of this study is largely descriptive. First, we investigate the prevalence of SR-IRRs among students at two large public predominantly White universities (PWIs), paying special attention to university location and racial/ethnic differences in SR-IRR engagement. Second, we explore how students’ self-identifications differ from researcher-defined categorizations that label as “IR/IE” all partnerships that cross either racial or Hispanic ethnic boundaries (RD-IRRs). Multivariate analyses examine the factors associated with SR-IRRs and RD-IRRs, as well as the extent to which race/ethnicity, university location, and nativity status interact to shape individuals’ engagement in relationships across racial/ethnic lines and their likelihood of self-identifying that engagement as “interracial.” Finally, we discuss the implications of our findings for our understanding of the prevalence and potential meaning of interracial relationships among this age group, as well as suggestions for future data collection efforts and avenues of research.
Methods
Data
We collected data in Fall 2015 and Spring 2016 at two large U.S. public 4-year universities, one located in the Southeast (n = 545) and one in the Midwest (n = 310). Respondents were recruited to participate in an anonymous online survey through blind email lists that were compiled using a random sample of the enrolled student bodies. Respondents were required to be at least 18 years old and have been in a romantic relationship within the past 12 months. Students were sent two follow-up email invitations, spaced over a period of 2 weeks. No incentives were offered for participation in the online survey, but the invitation did indicate that upon completion of the survey some students would be given an opportunity to volunteer for an in-person interview that would receive monetary compensation.
Although the findings of this study are not nationally representative, both subsamples are drawn from large public institutions (SE = ∼41,000 students; MW = ∼22,000 students) and the use of random sampling is an advance over much previous research on college-aged adults that relies primarily on convenience sampling (e.g., Field et al. 2013; Knox et al. 2000; Levin et al. 2007). For the multivariate analyses, the two subsamples of data are pooled together, a methodological decision backed by previous research on the appropriateness of combining multi-site survey data (Pearlson 2009; Zhou et al. 2017; Zhou et al. 2018) and in line with previously published work using this specific dataset (Miller et al. 2022). Doing so increases the overall sample size of the dataset which, in turn, gives the dataset sufficient statistical power to capture the nuanced differences in smaller populations, such as racial/ethnic minorities who are engaging in interracial relationships. It also allows us to test for statistically significant differences in outcomes and predictors across sites and ensures a large enough sample size to examine potential moderating relationships (Pearlson 2009; Zhou et al. 2017; Zhou et al. 2018). Analysis of the two university subsamples in separate multivariate models precludes these examinations, but does produce findings that follow very similar patterns to those derived using the larger pooled sample (additional results available upon request). Since we used surveys with standardized measures to collect the data simultaneously at both sites, our multi-site data collection methodology avoided problems related to post-hoc data pooling and site-specific effects on measures that limit some other multi-site studies (Zhou et al. 2018).
Overall, the full analytic sample includes 788 students who had complete information for all variables of interest (SE university = 489; MW university = 289). A total of 8 students, 5 from the SE university and 3 from the MW university, were dropped from the initial sample due to incomplete information. During the data collection process, no information was gathered regarding the specific combination of racial identities of those who reported being multiracial. Since we were unable to determine whether multiracial individuals and their partners had crossed any racial or ethnic boundaries, we also dropped the 69 students (51 SE university; 18 MW university) who identified as multiracial. Finally, given that there were only two Native American students, those individuals were also dropped from the final sample. We can estimate a crude response rate based upon how many emails were sent and the number of completed surveys. This estimated response rate (12% for the SE university, 6% for the MW university, and 9% overall) is a very conservative estimate, however, as the invitation made it clear that there were firm participation criteria that excluded many email recipients. The study was approved by both universities’ Institutional Review Boards, and the use of the email lists was approved by the universities’ Offices of Institutional Research.
Measures
To better understand how contemporary college students label their relationships, we examine both self-reported “interracial” relationship status and a researcher-defined measure of the concept. Given that the survey only asks questions about respondents’ most recent relationships, both of these measures provide conservative estimates of life-time experience with interracial romance.
Self-Reported Interracial Relationship (SR-IRR) is measured as a dichotomous variable using information from the following question, “Do you consider your current/most recent romantic relationship to be an interracial relationship?” This question was asked after respondents had already been queried about their partner’s race/ethnicity. Eleven other non-race-related questions about the respondents’ partner and their relationship together came between the two questions. Respondents were given only “yes” or “no” response options, and there were no missing values. While the structure of this question did not capture potential uncertainty in how individuals identify their relationships, it does show that college students will make a choice when given specific options.
Researcher-Defined Interracial/Interethnic Relationship (RD-IRR) is derived from the respondent’s self-reported race and Hispanic ethnicity and their report of the current/most recent partner’s race/ethnicity. The wording of the original survey question asked respondents to choose one racial/ethnic category with which they most closely identify (or most closely identify their partner) from the following six options: White, Black, Hispanic, Asian, Native American, and multiracial. In cases where the specific racial/ethnic designation of the respondent and their partner differs, the respondent is categorized as engaging in an RD-IRR. Thus, this measure captures relationships in which there is a racial difference as “traditionally” defined (i.e., crossing boundaries between White, Black, Asian, and Native American backgrounds), as well as relationships in which there is only a difference in Hispanic ethnicity.
Unfortunately, data limitations preclude us from capturing any non-Hispanic ethnic differences that might exist between partners (e.g., ethnic differences between two Asian partners) or any cultural differences that exist between partners who share a broad Hispanic background (e.g., differences between those of Cuban and Mexican descent). Given that we were unable to determine whether multiracial individuals and their partners had crossed any racial or ethnic boundaries, we removed multiracial respondents from the sample. Monoracial respondents who reported a multiracial partner, however, were retained and coded as being in an RD-IRR with their partner. We recognize that this is a limitation, as the two partners may share some portion of their racial and/or ethnic background.
University location
To determine the effect of location on both prevalence of RD-IRRs and the tendency to self-identify as being in an interracial relationship, we create a dichotomous indicator that specifies whether the respondent was currently enrolled in the Southeastern (SE) university or the Midwestern (MW) university.
Control variables
We develop several measures to control for factors potentially associated with both RD-IRR formation and self-identification behaviors. These include self-reported socio-demographic background characteristics, current/most recent relationship status, interaction with other-race friends, and interracial relationship exposure and attitudes among family and friends. Respondents reported Age in years, specified Gender as either male or female, and specified Nativity Status as either native-born or foreign-born. Racial/Ethnic Background was categorized as one of the following: non-Hispanic White, non-Hispanic Black, Hispanic, or Asian. Respondents reported Academic Status by indicating their current year in school. These responses were re-coded into a dichotomous measure indicating whether respondents are undergraduate or graduate/professional students. Current/Recent Relationship Status information pertains to the respondent’s current/most recent relationship and is coded into a series of four dummy variables: currently dating, currently cohabiting, currently married, and recent past relationship. Friends of Other Race is measured as a dichotomous variable that indicates whether the respondent has three or more close friends of a race/ethnicity different than their own (sample median = 2). Data was obtained through a direct question, “How many friends do you frequently have face-to-face interact with that are members of a different racial group than yourself?” (original response categories = 0, 1–2, 3–4, 5–6, and 7+). Family IR/IER Experience and Friend IR/IER Experience are measured as dichotomous variables that indicate whether any immediate family members or close friends had ever had an “interracial” romantic relationship. Friends Supportive of IR/IERs is a dichotomous variable that captures whether the respondent believes that, in general, their close friends would be supportive of them being in an “interracial” relationship. These questions about family and friend engagement in and attitudes about interracial relationships were asked after the respondent was questioned about how they identified their own relationship.
Analytic Strategy
All analyses are conducted using Stata-SE, version 17.0. Percentages, means, and standard deviations are used to describe the distribution of our study variables for the sample as a whole and separately by university location and self-reported relationship type (i.e., interracial or not). Descriptive tables then present a comparison of the percent of respondents who self-report an interracial relationship and the percent classified as being in an RD-IRR, as well as a breakdown of the particular racial/ethnic partnership combinations found among respondents in SR-IRRs, RD-IRRs, and in situations where respondents did not identify themselves as being in an interracial relationship but the researchers did categorize them as being in an RD-IRR. This latter group is further explored descriptively to determine which groups of respondents are most likely to identify their relationship as same-race, despite crossing a racial and/or Hispanic ethnic boundary in their partnership.
One of our primary research aims is to assess whether our understanding of the factors that predict interracial relationship engagement would differ depending upon how interracial relationships are measured (i.e., if we use reports of self-identification as opposed to the researcher-derived measures most commonly used in the literature). To do this, we employ separate logistic regression models to examine the predictors of SR-IRRs and the predictors of RD-IRRs. We then focus our examination on those respondents for whom self-identity does not match researcher-imposed definitions. In particular, we employ logistic regression models to develop a better understanding of how college students who choose not to self-identify as “interracial” a relationship that crosses a racial/ethnic boundary are distinct from their peers. Unfortunately, the number of individuals who self-identify themselves as being in an “interracial” relationship, but are not classified as crossing a racial/ethnic boundary by the RD-IRR measure, is too small for any meaningful examination in a multivariate context (n = 10). We use the margins command to estimate predicted probabilities for significant interactions.
Descriptive Results
Demographic Characteristics of Respondents, by University Location and Most Recent Relationship Type as Reported by Respondent.
Note. Significant differences in characteristics between those in self-reported interracial relationships and those in self-reported same-race relationships.
^p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001.
There are potentially important differences across university subsamples in terms of interaction with other-race friends, IR/IER exposure, and friends’ IR/IER attitudes. While a minority of the overall sample reports 3+ close friends of another race, a larger percentage do so at the SE university (36.3%) than the MW university (29.1%). A larger percentage of students at the SE university also report immediate family member and close friend experience with an interracial relationship (SE: 44.5% and 58.7% vs. MW: 40.1% and 54.3%) and that their friends would be supportive of IR/IER engagement (90.4% vs. 85.5%).
Table 1 also shows the prevalence of current/recent SR-IRRs, broken down by the respondents’ background characteristics. Of the 788 respondents in the full sample, 18% self-report that they are in an “interracial” relationship (MW: 19.4%; SE: 17.2%). SR-IRRs vary significantly by respondents’ race/ethnicity and nativity status, as well as their exposure to other race friends, family and friend IR/IERs, and friend support for IR/IERs. In the full sample, 66.7% of Asian and 29.7% of Hispanic respondents report an SR-IRR, compared to only 20.8% of non-Hispanic Black and 13.5% of non-Hispanic White respondents. In addition, foreign-born respondents are two to three times as likely to report an SR-IRR as native-born respondents. Most patterns of statistical significance are fairly similar across university locations, but we find greater prevalence of SR-IRRs in the Southeast for Asian students and in the Midwest for Hispanic and non-Hispanic Black students. Overall, these prevalence figures are lower than what we see in much of the previous research on college populations, which typically use a researcher-defined method of categorization. Across both universities, the prevalence of SR-IRRs is significantly higher among those who report having 3+ close friends of another race, having immediate family or close friends with IR/IER experience, and having close friends who would be supportive of IR/IER engagement.
Percent of Recent Relationships Classified as Interracial Using Different Criteria, Full Sample, N = 788.
*Note. Of those who self-reported having an "interracial" relationship, 7% were not classified as having an RD-IRR by the researchers (8 non-Hispanic White respondents with NH White partners and 2 Asian respondents with Asian partners).
**Of those who did not report their relationship as "interracial," 9.4% (35 non-Hispanic White, 2 Asian, and 24 Hispanic respondents) were classified as RD-IRR by the researchers. Of these 61 respondents, 9 reported a racial difference between themselves and their monoracial partner (1 Hispanic, 5 non-Hispanic White, and 2 Asian respondents). This made up 1.1% of the total sample. Another 14 respondents indicated that they were not in an SR-IRR and then reported that their partner was multiracial (2 Hispanic and 11 non-Hispanic White respondents; 1.8% of the total sample). The remainder (4.8% of total sample) indicated that they were not in an SR-IRR and then reported Hispanic ethnicity differences between themselves and their partner.
^ Significant difference in % self-reported and % researcher-defined (chi-square tests, p < 0.001).
The final two columns of Table 2 show the percent of respondents whose self-reported status does not “match” the status imposed by researcher-defined measures of cross-group relationships, as well as the particular direction of the mismatch. Here, we see that only a very small portion of respondents label themselves as being in an “interracial” relationship, but are not considered to have crossed racial/ethnic boundaries by the researcher-imposed definition: 6.1% of Asian and 1.3% of non-Hispanic White respondents fall into this situation. A larger portion of respondents, particularly Hispanic respondents, fall into the opposite situation, where they do not identify as being in an “interracial” relationship, but the researcher-defined measure does count them as having crossed salient group boundaries. Overall, 26.3% of Hispanic, 5.7% of non-Hispanic White, and 6.1% of Asian respondents fall into this situation. In general then, Asian and Hispanic respondents are most likely to be categorized into an interracial relationship, by either definition, and also appear to experience the highest degree of variability in how they think about the concept of race and self-identify their relationships.
% Classified as RD-IRR by Demographic Characteristics, Among Those Who do Not Self-Report an "Interracial Relationship" (n = 646).
Note. Significant differences in the likelihood of being categorized as IR/IER: ***p < 0.001.
Significant Group Differences in Likelihood of Not Self-Reporting an "Interracial Relationship" and Being Classified as RD-IRR, by University Location.
Together, these findings indicate that Hispanic students, and to a lesser extent their Asian and non-Hispanic White peers, tend to self-report their relationships as interracial less frequently than would researchers using the measures often found in recent literature. As a result, a large portion of young people whom researchers label as engaging in cross-boundary relationships that are equivalent to “interracial” relationships (i.e., they combine them together in one measure) may not consider themselves to be in relationships akin to “interracial.” Indeed, this may hold for over one-quarter of all Hispanic college students and appears to be significantly more likely for Hispanic students at the MW university. Given the larger number of Hispanic students at the SE university, however, there is still a larger number of Hispanic students at that school who are found in this situation.
Another important finding is that, regardless of definition used, we see no difference in reporting of cross-group relationships for non-Hispanic Black respondents. These students always label relationships that cross group boundaries, either racial or Hispanic ethnic, as being “interracial” in nature. Also, unlike a small number of Asian and non-Hispanic White respondents who report “interracial” relationships with a partner of their same racial background (and, presumably, a different intra-racial ethnic background), no non-Hispanic Black respondents report a relationship with a Black partner as being “interracial.” Thus, for non-Hispanic Black college students, within-racial-group ethnic distinctions (e.g., African American vs. Black Caribbean ethnicities) are never considered salient enough to override a shared Black identity.
Partner’s Race/Ethnicity Within Most Recent Relationship, by Respondent’s Own Race/Ethnicity and Relationship Type.
As previously noted, a small portion of the students report that they are in an SR-IRR, but they do not fit in the RD-IRR categorization. Of those self-reporting an SR-IRR, 9.6% of non-Hispanic Whites and 9.1% of Asians report having a partner who does not cross either racial or Hispanic ethnic boundaries. Location appears to matter, however. While only 5.6% of non-Hispanic Whites at the MW university fell into this situation, 12.8% of non-Hispanic Whites at the SE university did. Conversely, 0% of Asians at the SE university, but 22.2% of Asians at the MW university, were in this grouping. These findings suggest that non-Hispanic Whites at the SE school and Asians at the MW school may be more likely than their peers at the other location to experience partnerships with intra-racial ethnic or cultural differences (e.g., Chinese vs. Japanese or Korean and “American” vs. European, especially foreign-born European) that they view as indicative of boundaries worthy of labeling as “interracial.”
Overall, a larger portion of the students report that they are not in an interracial relationship, but have a partner who is of a different racial and/or Hispanic background. As indicated in Table 5, these cases disproportionately occur at the SE university (approximately 10% of SE respondents vs. 3% of MW respondents) and the vast majority are cases in which the partnerships are composed of either one Hispanic and one non-Hispanic White partner or one non-Hispanic White and one multiracial partner.
In sum, we see a great deal of variation in how college students, especially Hispanic students, label their romantic relationships. At the same time, there appears to be no variation in how non-Hispanic Black respondents define cross-group relationships. The findings for Black respondents are compelling and supportive of prior research that shows Black Americans to be a group still viewed as distinctly different from all other racial and ethnic groups (Bonilla-Silva 2004).
Multivariate Results
We use logistic regression to examine the factors associated with SR-IRRs, RD-IRRs, and the situation in which respondents do not identify themselves as being in an SR-IRR, but the RD-IR/IER categorization does. We also examine the extent to which university location, race/ethnicity, nativity status, and exposure to interracial relationships interact to shape individuals’ engagement in relationships across racial/ethnic lines. For ease of interpretation, we present figures that illustrate adjusted predicted probabilities for all significant interactions, calculated using the margins command in STATA version 17.
Logistic Regression Predicting SR-IRR, RD-IRR, and the Mis-Match of Not Reporting an SR-IRR but Being in an RD-IRR (N = 788), Odds Ratios and Standard Errors.
Note. ^p ≤ 0.10; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.
Model 3 shows significant interaction effects, which differ across the two outcomes. For SR-IRRs, the interaction between nativity status and university location is significant. As shown in Figure 1, the predicted probability of SR-IRR is significantly higher for foreign-born respondents at the SE university (p = 0.25) than the MW university (p = 0.12). The opposite holds true, however, for the native-born (SE p = 0.11; MW p = 0.19). For RD-IRRs, university location moderates the effect of race/ethnicity. Figure 2 shows that, for Hispanic and non-Hispanic Black individuals, the probability of engagement in an RD-IRR is significantly higher at the MW university than the SE university (Hispanic MW p = 0.77, SE p = 0.44; NH Black MW p = 0.31, SE p = 0.11), while the opposite is true for Asians (Asian MW p = 0.43, SE p = 0.73). University location matters little for the probability that non-Hispanic Whites’ relationships will cross group lines (MW p = 0.15, SE p = 0.18). These interactions suggest that university location matters for engagement in interracial relationships, but in different ways for the two measures of engagement and for different socio-demographic groups of students. Adjusted predictions of self-reported “interracial” romantic relationships, by nativity and university location (n = 788). Adjusted predictions of RD-IRR, by race/ethnicity and university location (n = 788).
The findings in the final panel in Table 6 show that the situation of having not self-reported a relationship as “interracial,” but being categorized as being in an RD-IRR, is predicted by different factors than are the two other measures of engagement. University location is strongly and consistently associated with this outcome, with those attending the SE university 3.18 times as likely as those attending the MW university to experience this situation (this reduced to 2.60 times as likely after controlling for background factors in Model 2). Model 2 indicates that, compared to their non-Hispanic White peers, Hispanic students are 4.45 times as likely to experience this situation. Conversely, non-Hispanic Blacks never experience this situation and, as a result, are dropped from this set of analyses. Model 2 also adds measures of current relationship status and exposure to and perceptions of approval for IR/IERs. Here, we see that the only factor even marginally associated with the likelihood of experiencing a mismatch between self-identification and researcher-identification is having a close friend with IR/IER experience. Those who have at least one friend with this type of relationship history have 40% lower odds.
In addition to having direct effects, Model 3 shows that race/ethnicity is moderated by both university location and whether or not respondents have a close friend who has had IR/IER experience. Figure 3 illustrates some striking racial/ethnic differences: non-Hispanic Black respondents are never in this situation and both Asian and non-Hispanic White respondents have very low probabilities regardless of location (Asian p = 0.05; NH White p = 0.01–0.08). Hispanic respondents’ probabilities, however, are much higher (p = 0.21–0.36). Thus, while Hispanic students are among the most likely to cross group boundaries for romance, particularly ethnic boundaries, they appear to have much greater flexibility than their peers in how they self-report those relationships. Moreover, compared to Hispanic students at the SE university, those at the MW university are both more likely to engage in cross-group relationships and more likely to choose not to self-report such relationships as “interracial.” Figure 4 presents the predicted probability of this outcome and highlights the significant interaction between race/ethnicity and having at least one close friend with IR/IER experience. As this figure shows, Hispanics are the only group for whom having friends with IRR experience is associated with their likelihood of not self-identifying a boundary-crossing relationship as “interracial.” Indeed, Hispanic students who have friends with IR/IER experience have three times lower probability of doing so (with friends p = 0.10; without friends p = 0.32). Adjusted predictions of not being in a self-reported “interracial” relationship, but being categorized in a RD-IRR, by race/ethnicity and university location (n = 788). Adjusted predictions of not being in a self-reported “interracial” relationship, but being categorized in a RD-IRR, by race/ethnicity and close friend experience with interracial or interethnic relationships (n = 788).
Discussion and Conclusions
Using data collected through random sampling at two large public 4-year universities, one in the Midwestern (MW) and one in the Southeastern (SE) region of the U.S., we document the prevalence of SR-IRRs among college students and explore the extent to which self-identifications differ from researcher-defined categorizations that label all partnerships that cross either racial or Hispanic ethnic boundaries as “interracial/interethnic” (IR/IERs). We also examine the factors associated with both SR-IRRs and researcher-defined IR/IERs, as well as the extent to which race/ethnicity, university location, and nativity status interact to shape individuals’ engagement in cross-group relationships and their likelihood of self-identifying that engagement as “interracial” in nature.
When students are asked whether their current/most recent relationships are “interracial,” without having that term defined for them, roughly 18% at both universities in this sample indicate that they are. Self-reported IRRs are more common among the foreign-born, Hispanics, and Asians and least common among non-Hispanic Whites. Patterns vary by university location, however, with a greater likelihood found among the foreign-born and Asian students in the Southeast and Hispanic and non-Hispanic Black students in the Midwest. Thus, in general, students from racial/ethnic minority backgrounds are more likely to self-report an IRR when their group makes up a smaller portion of their university’s population. Moreover, comparing data across the two university subsamples, we find that particular racial/ethnic minority groups are more likely to engage in SR-IRRs at the university in which that group is relatively smaller. These findings are largely in line with prior research on young people’s relationship behavior (Fujino 1997; Levin et al. 2007; Miller 2014; Tillman and Miller 2017; Perry 2013, 2014; Wang et al. 2006; Yancey 2002) and supportive of expectations derived from the literature studies on racial identity development, social distance, and intergroup contact.
Our findings also show substantial variation in how young people interpret the term “interracial” when they are answering this type of question. Some students appear to think that “interracial” relationships include only those that cross traditionally defined racial categories (i.e., White, Black, and Asian), not ethnic categories. Roughly 26% of Hispanics and 6% of non-Hispanic Whites who say that they are not in an “interracial” relationship go on to identify a partner of a different Hispanic ethnic background. Other students appear to define as “interracial” all relationships that cross any racial or ethnic boundary. The most common self-reported IRR combination at the SE university, for example, includes one non-Hispanic White and one Hispanic partner (31.4% of SR-IRRs). These varying interpretations of the term “interracial” reflect a lack of consensus on the meaning of race and ethnicity among a generation of youth raised in an increasingly diverse nation, with a quickly growing Hispanic population (Perlmann and Waters 2002). These varying interpretations may also reflect, as posited by Bonilla-Silva and colleagues (2000; 2008), a slowly rising conceptualization of Hispanics and Asians as “honorary Whites.” Clearly seen among the responses at both universities, however, is the enduring pattern of “Black exceptionalism” highlighted by scholars such as Lee and Bean (2010; 2012). Black students make up a group consistently considered distinct from all other racial and ethnic groups.
Comparing respondents’ self-reported information with a researcher-defined categorization of IR/IER illustrates how much variations in interpretation can influence our understanding of cross-group relationships among college students. Whereas 18% of respondents self-reported their relationships as “interracial,” if researchers use a definition that captures all racial and ethnic differences, over 24% of respondents are labeled as engaging in cross-group relationships. The magnitude of the difference that results from shifting definitions varies by respondent race/ethnicity, however, with Hispanics experiencing the most dramatic differences (26 percentage points) and non-Hispanic Black respondents experiencing no differences at all. Clearly, how we define, or leave undefined, the term “interracial” in survey questions significantly influences the information we gather, particularly for Hispanic students.
The use of researcher-defined measures that combine into one category respondents who consider themselves to be in an interracial relationship with those who do not may also muddy our understanding of how engagement in an IR/IER is related to other social and emotional outcomes for young people today. Those who do not view racial/ethnic differences with their partner as salient to how they define their relationship may experience fewer socio-emotional effects as a result of their relationship’s racial/ethnic composition than those who do, leading research to underestimate the significance of racial/ethnic composition for some groups’ outcomes. The lack of consensus on when a relationship constitutes an “interracial” relationship also holds implications for how researchers interpret the results of attitudinal surveys that assess respondents’ attitudes toward “interracial” dating and marriage.
Multivariate analyses help to further elucidate the factors associated with cross-boundary relationships. In general, similar factors are associated with both SR-IRRs and RD-IR/IERs, with race/ethnicity, friendship diversity, and friend/family experience with IRRs significantly predicting both outcomes. Although university location does not directly influence engagement in cross-group relationships, net of other controls, location significantly interacts with nativity status in predicting SR-IRRs and with race/ethnicity in predicting RD-IR/IERs.
We also used multivariate analyses to examine the factors that predict being in the group that does not report an SR-IRR, despite crossing a racial/ethnic boundary in their relationship. Here, we see not only the continuing effect of race/ethnicity (particularly being Hispanic) and friend experience with IRRs but also a powerful direct association between this outcome and university location, such that falling into this group is significantly more likely at the SE university than the MW university. Again, however, location interacts significantly with race/ethnicity in important ways. For Hispanics, the students with the highest likelihood of being in this group, attending the MW university is actually associated with a higher predicted probability (MW = .36 vs. SE = .21). So, overall, Hispanic students are among the most likely to cross racial/ethnic lines for romance (mostly ethnic lines), are more likely to do so at the MW university where they are a smaller portion of the overall student body, and are less likely to report their cross-group relationships (mostly with non-Hispanic White partners) as “interracial,” particularly at the MW university. Hispanic students whose relationships cross racial/ethnic lines are less likely to say that they are not in an SR-IRR, however, if they have at least one close friend who has been in a cross-group relationship. Future research should continue to disentangle the importance of race/ethnicity and location and strive to replicate these findings in other geographic regions and among students at other types of educational institutions.
Overall, these findings have important implications for methodology and measurement, as well as the substantive field of study. If researchers who focus on young adults are interested in studying romantic relationships that cross racial and ethnic group lines, they must be very deliberate in how they ask respondents about their relationships. Surveys and interview protocols should ask specific questions about the racial and ethnic backgrounds of all the individuals involved and also ask respondents how they personally label their relationships, clearly defining for them the terms “interracial” and/or “interethnic.” Failure to do so may undermine our ability to understand how young people are thinking about and experiencing their cross-group relationships and keep us from a nuanced understanding of how IRRs affect individual and couple outcomes. Ultimately, it may also inhibit our ability to track changes in relationship behavior over time, as we continue to see shifts in young Americans’ ideas about race, ethnicity, and difference. Our findings suggest that all of these difficulties will be particularly pronounced for researchers interested in studying the relationship behaviors and socio-emotional outcomes of young adults in the growing Hispanic population.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.