Relationship Involvement Among Young Adults: Are Asian American Men an Exceptional Case?

Abstract

Asian American men and women have been largely neglected in previous studies of romantic relationship formation and status. Using data from the first and fourth waves of the National Longitudinal Study of Adolescent to Adult Health (Add Health), we examine romantic and sexual involvement among young adults, most of who were between the ages of 25 to 32 (N=11,555). Drawing from explanations that focus on structural and cultural elements as well as racial hierarchies, we examine the factors that promote and impede involvement in romantic/sexual relationships. We use logistic regression to model current involvement of men and women separately and find, with the exception of Filipino men, Asian men are significantly less likely than white men to be currently involved with a romantic partner, even after controlling for a wide array of characteristics. Our results suggest that the racial hierarchy framework best explains lower likelihood of involvement among Asian American men.

INTRODUCTION

While social scientists have examined patterns of marriage and cohabitation in young adulthood, they have paid less attention to “dating” (i.e., romantic involvement outside of a co-residential relationship) in this period (for exceptions see Blackwell and Lichter 2004; Sassler and Joyner 2012; Keels and Harris 2014; McClintock 2010). Studies that focus on population-based samples are a case in point. While a growing number of studies utilizing these samples have included romantic relationships along with co-residential unions, they have predominately focused on the racial mix of partners in relationships (Blackwell and Lichter 2004; Sassler and Joyner 2012). Studies highlighting differences in patterns of romantic involvement among racial and ethnic groups have focused largely on adolescents (e.g., O’Sullivan, Cheng, Harris, and Brooks-Gunn 2007; Raley and Sullivan 2010). Prior studies on the relationship patterns of young adults have not given adequate attention to how partnering differs across race and ethnic groups (Sassler 2010). Especially troubling is the fact that the vast majority of these studies fail to distinguish the relationship patterns of Asian American men and women.

In a recent exception, Brown, Van Hook, and Glick (2008) compared non-Hispanic whites, non-Hispanic blacks, Hispanics, and Asians (ages 20 to 34) with respect to current co-residential involvement and found that black men and women had the lowest levels of co-residential involvement (31.7% and 35.1%) compared to whites (47.4% and 65.8%), while Hispanics of Mexican origin had the highest (48% and 68.3%). Asians on the other hand, displayed the largest gender gap in co-residential involvement; just 35.8% of Asian men were either married or cohabiting, compared with 59.1% of Asian women. The gender gap in involvement among Asians is consistent with research highlighting the emasculation of Asian American men as well studies that document their marginalization from internet mate markets. For example, using a non-random sample of opposite-sex daters on Yahoo personals, Feliciano, Robnett, and Komaie (2009) found that Asian American men were systematically excluded from by women from all racial and ethnic groups (including Asian American women). The focus on co-residential unions not only neglects those in romantic and sexual relationships, but also those who are excluded from romantic involvement.

We fill an important gap in research on young adult relationships by focusing on the experiences of Asian American men relative to patterns of romantic involvement for all major racial, ethnic, and gender groups simultaneously. Data from the most recent (i.e., fourth) wave of the National Longitudinal Study of Adult and Adolescent Health (Add Health) offers us the opportunity to reassess patterns of current romantic involvement among young adults using a broader definition of relationship and a more exhaustive set of race/ethnic groups (i.e., whites, blacks, Hispanics, and Asians) than those used by prior studies on this topic. We first review and synthesize competing frameworks concerning factors that promote and impede relationship involvement among young adults. We focus our attention on the implications of these frameworks for Asian Americans, a group that is generally neglected by studies focusing on union formation and status. Next, we present descriptive statistics on current involvement in different-sex romantic/sexual relationships for all four race/ethnic groups of men and women based on reports from the Wave IV interviews, conducted when most respondents were between the ages of 25 and 32. We then consider the role that individual characteristics play in race-ethnic gaps among men and women using logistic regression models. Importantly, our study is among the first to focus on any relationship involvement among a national sample of young adults across multiple racial and ethnic groups and by gender in examining patterns of exclusion from the dating market.

BACKGROUND

Theoretical frameworks addressing romantic partnering have typically focused on the decision to form a romantic relationship, preferences for characteristics of partners, and goals for relationships (e.g., sex and marriage). According to these frameworks, aspects of partnering not only differ by period of the life course, but also across race and ethnic groups (Sassler 2010). Past research has used both structural and cultural explanations to understand race and ethnic differences in union formation (Raley and Sweeney 2009), and here we extend this work to examine current romantic/sexual involvement during early adulthood. We ask the question, why are some adults at this stage of the life course not currently partnered? Further, we add to the literature by considering a third explanation that emphasizes the role of racial hierarchies in shaping romantic involvement.

Structural explanations suggest that the ability to form a romantic relationship depends on how individual-level sociodemographic and other traits (e.g., earnings potential and physical attractiveness) are valued in the broader mate market (England and Farkas 1986). It may be that those who are not currently involved simply lack both economic and physical resources necessary to form a romantic relationship. In addition, structural explanations highlight the role that imbalances in local sex ratios (e.g., the number of men per 100 women in a metropolitan area) in shaping the availability of potential romantic and sexual partners (Fossett and Kiecolt 1991). In contrast, cultural explanations suggest that the norms and values of some racial and ethnic groups (e.g., strict parental control over dating, obligation to the family) impede their romantic relationship involvement (Schneider and Lee 1990; Smith 2006; Espiritu 2001). Critical race perspectives explicitly argue that racial hierarchies define desirability in ways (e.g., the equation of female attractiveness with white standards of beauty) that marginalize some groups of men and women in broader mate markets (Burton et al. 2010). We elaborate on all three of these perspectives below to develop expectations on racial and ethnic patterns of romantic involvement among men and women, but focus on Asian Americans.

Structural Explanations

Individual Characteristics

Structural frameworks highlight the role of individual characteristics in explaining the formation of romantic relationships. Drawing heavily from theories of exchange (England and Farkas 1986), prior research suggests that partner’s trade on their personal characteristics (e.g., wealth, education, physical attractiveness) in the mate market (Blackwell and Lichter 2004; Lichter 1990). The assumption of this framework is that individuals seek partners with the most desirable characteristics and the characteristics desired in a mate are gendered. For example, physical attractiveness has long been valued in women while economic resources have traditionally been valued in men (Sassler and Joyner 2011). As individuals who are most desirable pair off with each other, those who are less desirable end up with partners who are comparable in terms of desirability (Becker 1981; England et al. 1986; Fu 2001; Qian and Lichter 2007; Qian 1997; Rosenfeld 2005; Schwartz and Mare 2005) or end up unpartnered altogether (Cawley, Joyner, and Sobal 2006). Studies continue to show that employment and earnings increase the likelihood of marriage, especially among men (e.g., Xie, Raymo, Goyette and Thornton 2007). However, prior research suggests that gender differences in the role of economic factors have declined in recent decades as economic prospects have become more important for women’s marriage prospects but have remained critical for men (Sweeney 2002).

Assuming that higher socioeconomic attainment of men increases their desirability as potential mates, we would expect Asian American men to have better prospects of involvement than their white, black, or Hispanic male counterparts. After all, they have much higher levels of educational attainment and income. Kao, Vaquera, and Goyette (2013) find that among adults 25 and older, almost 50% of Asian Americans have a bachelor’s degree or higher, compared to about 30% of whites, 18% of blacks, and 13% of Hispanics. Still, there is considerable heterogeneity in the educational attainment of Asians, with South Asians, Chinese, and Koreans faring better than whites and those from Cambodia and Laos doing worse (Kao and Thompson 2003). Asian women too should fare better than their counterparts, but as suggested earlier, women’s prospects for marriage are less strongly linked with their economic prospects. Still, it remains an empirical question whether Asian men and women witness similar “relationship-returns” for education as their white counterparts.

Prior studies have examined the role of physical characteristics in mate selection, but they have most heavily focused on associations between partner characteristics in marriage (e.g., Udry 1977). A recent spate of studies, however, has explored how physical characteristics are linked to current romantic involvement using population-based samples (e.g., Cawley et al 2006; Manfredini et al. 2012). Generally speaking, these studies suggest that physical attractiveness is more strongly linked to involvement for women than for men. Research also reveals a preference for the male partner to be taller than the female partner in a relationship and emphasizes that average height varies with ethnicity but is correlated across genders within the racial groups. The fact that Asians are, on average, shorter than whites and blacks, may account for some of the disadvantage in the dating market faced by Asian American men (Belot and Fidrmuc 2010).

Cultural Explanations

A cultural framework suggests that some racial and ethnic groups, particularly those composed of a high proportion of immigrants, may have distinctive patterns of involvement due to their family values, such an emphasis on educational achievement (Glick, Ruf, Goldscheider and White, 2006; Schneider and Lee 1990), cultural views about dating and premarital sexual behavior (Espiritu 2001) and a heightened sense of obligation to the family (Smith 2006). Research that relies on a cultural framework to explain Asian patterns of union formation stress the influence of family ties that places strong constraint on obedience to parents which may limit premarital sexual behavior and cohabitation (Chan 1994). Research by Espiritu (2001) supports this notion, finding that Filipino first-generation parents retain strict control over their daughters’ dating choices as compared to their sons. Moreover, the control of parents over dating ought to affect women more than men, so to inhibit the relationship formation of Asian American women more than Asian American men.

Similarly, researchers argue that close ties to parents may distinguish patterns of relationship involvement for Hispanics. For instance, Valenzuela (1999) suggests that familism (which is proxied by the time spent with family and the importance of helping family members) may be more important to Hispanic families than non-Hispanic families. Other research on Hispanic families suggests a strong need for parental approval of romantic relationships among youth as well as stronger family control over daughters’ dating behavior than sons (Torres 1998). Given their stronger family ties, we would expect that both Asian and Hispanic young adults would delay romantic partnering in comparison to white youth. If cultural values mattered, we would expect traditional “Asian” or “Hispanic” cultural values to produce gender differences in romantic partnering among Hispanics and Asians such that more conservative gender roles dictate that romantic involvement is less acceptable for women than for men. In other words, if immigrant Asian parents or Hispanic parents exercised greater control over their children, they would exercise greater restrictions on women than men.

At the same time, cultural explanations of whether a young adult chooses to be in a romantic relationship may become muddied given the disparity in cultural traditions among Asians and Hispanics. It is widely recognized that Hispanic and Asian ethnic groups differ markedly with respect to immigration histories and socioeconomic position (Kao, Vaquera, and Goyette, 2013; Bean and Tienda 1987), and also with respect to family formation patterns. For example, Puerto Ricans exhibit lower rates of marriage than any other racial or ethnic group in the US (Landale and Fennelly 1992) compared with those of Cuban origin that have higher rates of marriage than other groups (Landale, Oropesa, and Bradatan 2006). In addition, we know very little about heterogeneity within the Asian American population.

Further complicating these explanations, cultural factors may interact with structural factors to produce different patterns of relationship involvement for youth from minority families (Wilson 2009). For example, parental attitudes about dating and union formation may be intertwined with their aspirations for their children’s educational attainment (Cheng and Landale 2011). Schneider and Lee (1990) suggest that East Asians see their educational achievement in terms of what it means for the family rather than as an indicator of individual success, and are socialized to think in terms of filial piety, putting their duties to their parents above all else. Indeed, Cheng et al. (2011) finds that educational investments contribute to lower odds of early co-residential union formation among Asian Americans compared with whites. However, this research is limited because of a focus on early union formation (prior to age 25), a narrow definition of relationship involvement (i.e., cohabitation or marriage), and a lack of attention to possible gender differences. While prior research suggests that co-residential involvement is delayed by educational pursuits (Thornton, Axinn, Teachman 1995), it is unknown whether the formation of any romantic relationship is also delayed or whether this may vary by race and ethnicity.

Prior Romantic Involvement

Offering support for the cultural framework, prior research finds that romantic relationship behaviors among adolescents differ markedly by race and ethnicity (O’Sullivan et al. 2007). Given that adolescent romantic relationships are an important precursor of union formation in early adulthood (Raley, Crissey, and Muller 2007; Thornton, Axinn, and Xie 2007), we may expect racial and ethnic patterns of adult romantic partnerships to mirror adolescent patterns. Using data from the Add Health, Carver, Joyner, and Udry (2003) document how black, white, Hispanic, and Asian youth (most of who are between the ages of 12 and 18) differ with respect to romantic involvement in the eighteen months prior to the interview. They find that Asians are less likely to report romantic involvement than whites, blacks, and Hispanics, who have roughly similar levels of involvement in adolescence. They also suggest that gender differences within race/ethnic groups are minimal. More recent studies using the Add Health have investigated the link between adolescent romantic involvement and union formation behaviors in early adulthood (e.g., Raley, Crissey, and Muller 2007), finding continuity between romantic involvement during adolescence and the likelihood of co-residential union formation among young adults. Yet, this research is limited because it focuses on relatively early union formation (using the third wave of the Add Health when most respondents were between the ages of 18 and 25), and it does not examine variations in outcomes by race and ethnicity (for an exception see Cheng et al. 2011). It is less apparent whether the race and ethnic patterns of relationship involvement among adults will mirror those among adolescents.

Racial Hierarchies

Critical race perspectives focus on how certain race and gender groups are favored or marginalized in the mate market. In other words, the ability of an individual to enter into a romantic relationship may be hampered by set of perceived or ascribed differences attributed to their racial or ethnic group (Burton et al. 2010). Studies have suggested that unflattering stereotypical media depictions of nonwhites have contributed to a racial hierarchy in many aspects of society, including mate preferences (Bonilla-Silva 2010; Larson 2006). For example, media studies continue to document the racialized portrayals of Asian men’s masculinity as desexualized or effeminate (Feng 2002; Eng 2001) and black women’s femininity as less than desirable (Larsen 2006; Collins 2004; Wallace 1990). Scholars suggest that a preference for white standards of beauty reduces black women’s opportunities to date or intermarry outside of their race (Collins 2004; Bany, Robnett and Feliciano 2014). Other research, however, highlights the preferences of black women, noting they hold the least favorable attitudes toward selecting a partner of a different race (Davis and Smith 1991; Todd, McKinney, Harris, Chadderton and Small 1992) and are the least likely to intermarry or date across race because of cultural influences and lack of trust toward non-Hispanic whites (Childs 2005).

Research on dating preferences provides additional evidence of this racial hierarchy. A recent study of internet daters finds that among those who expressed a racial preference, less than 10% of Asian men would not consider dating Asian women, yet approximately 40% of Asian women would rule out dating Asian men. It also reveals that more than 90% of women of all different racial groups who expressed a racial preference excluded Asian American men. In addition, men of all different racial groups are most likely to exclude black women than any other women (Feliciano, Robnett and Komaie 2009).

This gendered gap in dating preferences and experience is not only reflective of a racial hierarchy but is also mirrored in existing patterns of interracial romantic relationships. For example, prior research has documented large gender differences in interracial relationships formation among blacks and Asians. Specifically, research finds that black women and Asian men are much less likely than their same-race counterparts (i.e., black men and Asian women) to be coupled with a different-race partner (Qian 1997; Qian and Lichter 2007; Qian and Lichter 2011). A 2005 Gallup Poll found 9% of all women compared to 28% of all men said that they had dated an Asian. This included 9% of white women (compared to 27% of white men), 3% of black women (compared to 24% of black men) and 8% of Hispanic women (compared to 31% of Hispanic men) who had ever dated an Asian. Due to sample size restrictions, the report did not specify the dating patterns of Asian respondents. In more recent estimates, 17% of black newlyweds married someone who was not black, but the share of out-marriage among black women was less than half that of black men, 9% compared to 24%, respectively. Among Asians, just 17% of male newlyweds married outside their race compared with 36% of female newlyweds (Wang 2012).

Expectations for Asian Americans

The structural, cultural and racial hierarchy explanations presented above suggests several differences between Asian Americans and other race/ethnic groups. First, the structural explanation posits that based on a position of socioeconomic advantage, Asian American men and women will have higher levels of current involvement than any other race/ethnic group. Prior research also suggests that physical attributes such as height for men and attractiveness for women will be associated with higher levels of partnering. Second, cultural explanations for race and ethnic differences in partnering suggest that Asian Americans will be less likely to form romantic partnerships due to their own cultural preferences. If this is the case, we would find that Asian Americans are less likely to form romantic partnerships than whites, and that Asian American women would be less likely than Asian American men to form romantic partnerships due to their parents’ more traditional beliefs about gender roles. The same argument could be applied to Latinos. Further, these perspectives suggest that the delays into romantic unions as adolescents will continue into early adulthood for Asian American men and women. Finally, a racial hierarchy explanation suggests that Asian American men will be less likely than Asian American women to be partnered, as Asian American men face gendered cultural stereotypes barring them from entry into romantic partnerships.

METHOD

Data and Sample

This project uses information from the first and fourth waves of the National Longitudinal Study of Adolescent to Adult Health (Add Health) to examine differences by gender and race/ethnicity in relationship involvement among young adults. Add Health is a longitudinal school-based study. Using rosters from each school, Add Health selected a nationally representative sample of 20,745 adolescents in grades seven to twelve to participate in the first in-home interview. Add Health additionally selected oversamples of four race/ethnic groups: 1,038 black adolescents from well-educated families, 334 Chinese adolescents, 450 Cuban adolescents, and 437 Puerto Rican adolescents. The first in-home interview was conducted between April and December of 1995. The response rate for the in-home sample was 79%. In 2007 and 2008, the project conducted a fourth wave of in-home interviews for 15,701 of the original 20,745 respondents (a retention rate of over 75%). By the time of the fourth in-home interview, respondents were between the ages of 24 and 32. Importantly, Add Health used state-of-the-art survey methods to identify the romantic and sexual involvement of respondents, as well as their sexual orientation (i.e., computer-assisted self-interviews and partner rosters).

The sample is restricted to 5,840 males and 5,715 females. We began with 14,797 respondents who completed the Wave I and IV in-home interviews and had variables used to adjust for design effects (e.g., sample weights). Excluding respondents whose biological sex classification (marked by the interviewer) differed across Waves I and IV reduced the sample to 14,782. Of these respondents, 13,869 were classified as a single race/ethnicity: non-Hispanic white, non-Hispanic black, Hispanic, or non-Hispanic Asian. For conceptual clarity, we further restricted the sample to respondents whose most recent romantic/sexual partner was of a different sex, and who failed to report having any same-sex partners within the last year, resulting in a sample size of 11,717. Limiting the sample to respondents with information on key variables (e.g., height and interview-rated attractiveness) reduced the sample further to 11,555 respondents. We used survey procedures in Stata to take into account the complex design of Add Health; these procedures apply a post-stratification sample weight and identify participants geographically clustered within the same school or region of the country.

Variables

Our analyses utilize an indicator of different-sex partnering based on the Wave IV interview. We construct our dependent variable as a dichotomous indicator of any romantic partnering (i.e., dating, cohabiting or married) versus unpartnered. We divide respondents into mutually exclusive categories on the basis of their answers to questions on race and Hispanic descent at the Wave I interview: Hispanic (of any race), and non-Hispanic black, non-Hispanic Asian, and non-Hispanic white (the reference category). Non-Hispanic respondents who report more than one race were asked what category that best described their race and classified accordingly. We also include an indicator of nativity status (foreign born vs US born). Where possible, we break Asian and Hispanic respondents into ethnic subgroups. Using data from the first interview, we also construct measures of family structure (living with both biological or adoptive parents) and parental socioeconomic status (SES); the parental SES variable, which ranges from one to ten, and is based on the occupation and education of parents (Bearman and Moody 2004). (Our parental SES variable draws information from the parent and in-school questionnaires when missing information is missing from the in-home questionnaire.) We also include in our analyses measures typically included in studies of cohabiting and marital status or timing: age, and educational attainment by age 25. Educational attainment is measured as a single indicator of completion of a four-year college degree by age 25. In additional analyses, we include current college enrollment at the fourth interview.

In addition to these demographic variables, we also include less commonly used variables that capture opportunities and experiences forming romantic and sexual relationships. Based on the Wave I interview, we include a dichotomous variable of whether the respondent was involved in an adolescent romantic relationship (i.e., a relationship that the respondent defined as romantic or one that involved romantic activities) at any point in the eighteen months that preceded the interview. We also use data from the first interview to create a scale of attractiveness (ranging from 3 to 15) based on the interviewer’s rating of respondent’s physical attractiveness, personality attractiveness, and grooming (alpha = 0.767). The interviewer recorded the height of respondents at Wave IV, enabling us include their height in inches. We do not include measures of height from Wave I for several reasons: height was self-reported and it changed considerably from Wave I to Wave IV.

Analysis Plan

We begin this study by contrasting men and women who are white, black, Hispanic, and Asian according to their current relationship status. We pay close attention to how patterns of relationship status differ by race/ethnicity and sex. Again the sample is restricted to respondents considered to be seeking a different-sex partner. We also compare men and women of different racial and ethnic groups with respect to their values on the independent variables to better understand how they differ with respect to a constellation of factors likely correlated with romantic involvement. Next, we present three sets of logistic regression models (estimated for men and women separately) that address how current romantic involvement is associated with race/ethnicity and all other independent variables in both zero-order and full models. The first set of models addresses the structural explanation for race and ethnic patterns in current romantic involvement, whereas the second and third address the cultural explanations. While we do not have direct measures of cultural characteristics, we do include measures of nativity status and ethnic origin (i.e., Chinese, Filipino, Mexican, etc.).Taken together, analyses based on the descriptive statistics and models offer clues as to which factors potentially explain some of the racial and ethnic disparities in partnering at this stage of the life course.

RESULTS

Tables 1 and 2 display descriptive statistics (survey-adjusted means and proportions) for our sample of male (N=5,840) and female (N=5,715) young adults at the time of the Wave IV interview. These analyses are restricted to respondents who identify their last relationship as heterosexual and fail to report any recent sex or romance with some of the same-sex. Roughly one-half of respondents in the sample are female but the numbers of minorities are greater than in the general population, reflecting the oversamples of race/ethnic groups.

Table 1.

Means and Proportions (95% CI) on Variables by Race/Ethnicity, MALE (N=5,840)

	White		Black		Hispanic		Asian
Variable	Mean	95%CI	Mean	95%CI	Mean	95%CI	Mean	95%CI
Currently in a relationship (Wave 4)	0.80	(.77 -.81)	0.75	(.68 -.80)	0.78	(.73 -.82)	0.65	(.57 -.73)
Hispanic Ethnic Group
Mexican					0.62	(.48 -.75)
Cuban					0.05	(−.0 -.09)
Puerto Rican					0.11	(.05 -.16)
Central/South American					0.12	(.05 -.19)
Other Hispanic					0.10	(.05 -.13)
Asian Ethnic Group
Chinese							0.17	(.03 -.31)
Filipino							0.37	(.16 -.57)
Other Asian							0.46	(.27 -.64)
Wave 1 Variables
Parental SES	5.87	(5.6 -6.1)	5.15	(4.7 -5.5)	4.11	(3.8 -4.42)	5.80	(5.0 -6.5)
Living with both biological parents	0.69	(.66 -.71)	0.33	(.27 -.38)	0.83	(.74 -.91)	0.63	(.56 -.69)
IR attractiveness	10.40	(10.3 -10.5)	10.31	(10. -10.6)	10.42	(10.1 -10.7)	10.59	(10.3 -10.8)
Romantic relationship	0.65	(.62 -.67)	0.66	(.60 -.71)	0.61	(.55 -.67)	0.43	(.33 -.53)
Foreign Born	0.01	(.00 -.01)	0.01	(.00 - .02)	0.25	(.16 -.32)	0.50	(.35 -.64)
Wave 4 Variables
Age (in years)	28.41	(28.1 -28.6)	28.74	(28.3 -29.2)	28.47	(28. -28.9)	28.71	(28 -29)
Height (in inches)	70.36	(70.2 -70.5)	69.97	(69.7 -70.2)	68.57	(68.2 -68.9)	67.56	(67. -68.0)
College graduate prior to age 25	0.21	(.17 -.24)	0.11	(.07 -.14)	0.09	(.06 -.11)	0.43	(.28 -.56)
Number of Cases	3,402		1,164		920		354

Notes: Weighted estimates and unweighted sample sizes shown.

Note: All models are of current involvement with a different-sex partner among respondents who identify as 100% straight and report no involvement with a same-sex partner.

Table 2.

Means and Proportions (95% CI) on Variables by Race/Ethnicity, FEMALE (N=5,715)

	White		Black		Hispanic		Asian
Variable	Mean	95%CI	Mean	95%CI	Mean	95%CI	Mean	95%CI
Currently in a relationship (Wave 4)	0.84	(.82 -.86)	0.78	(.74 -.80)	0.82	(.78 -.86)	0.82	(.72 -.91)
HispanicEthnic Group
Mexican					0.57	(.42 -.71)
Cuban					0.07	(−.0 -.13)
Puerto Rican					0.10	(.05 -.15)
Central/South American					0.15	(.08 -.21)
Other Hispanic					0.11	(.07 -.15)
Asian Ethnic Group
Chinese							0.13	(.03 -.23)
Filipino							0.35	(.15 -.53)
Other Asian							0.52	(.36 -.66)
Wave 1 Variables
Parental SES	5.83	(5.5 -6.0)	4.63	(4.3 -4.9)	4.14	(3.8 -4.45)	5.41	(4.8 -5.9)
Living with both biological parϵ	0.67	(.63 -.69)	0.34	(.29 -.38)	0.80	(.70 -.89)	0.60	(.51 -.68)
IR attractiveness	11.19	(11.1 -11.3)	10.50	(10.3 -10.7)	10.88	(10.6 -11.1)	11.05	(10.6 -11.4)
Romantic relationship	0.69	(.65 -.72)	0.62	(.56 -.68)	0.53	(.48 -.58)	0.42	(.29 -.54)
Foreign Born	0.01	(.00 -.01)	0.02	(.00 - .02)	0.26	(.18 -.32)	0.57	(.45 -.68)
Wave 4 Variables
Age (in years)	28.26	(28. -28.5)	28.50	(28.1 -28.9)	28.41	(27.9 -28.9)	28.86	(28 -30)
Height (in inches)	64.78	(64.6 -64.9)	64.57	(64.3 -64.8)	63.00	(62.7 -63.3)	62.02	(61.6 -62.4)
College graduate prior to age 2	0.30	(.26 -.34)	0.19	(.14 -.23)	0.13	(.09 -.17)	0.37	(.21 -.51)
Number of Cases	3,163		1,325		925		302

Notes: Weighted estimates and unweighted sample sizes shown.

Note: All models are of current involvement with a different-sex partner among respondents who identify as 100% straight and report no involvement with a same-sex partner.

Among men (Table 1), there are clear gaps in romantic relationship involvement. While over three-quarters of white, black and Hispanic men report being in a relationship, just 65% of Asian men report any romantic partnership. We highlight variables for which there are significant gaps between men of different racial and ethnic groups. As documented in other studies, white and Asian men have higher means and percentages than black and Hispanic men on both indicators of SES: parental SES and educational attainment. However, it is important to note that Asian men are almost twice as likely as white men to possess a bachelor’s degree by age 25. Asian and Hispanic men are more likely to be foreign born than white or black men. Patterns by race for physical characteristics are less consistent. The interviewers at Wave I tend not to discriminate in rating the grooming and attractiveness of men from different racial and ethnic groups, as indicated by roughly comparable means in rating across the different groups. Asian men are considerably shorter than their counterparts. White men and black men are tallest (on average, 5′9″), followed by Hispanic men (5′7″), and Asian men (5′6″). Consistent with the findings of previous studies, Asian men are the group least likely to be romantically involved in adolescence.

Turning to Table 2, we find few unadjusted differences in the likelihood of any current romantic partnering for women of different race and ethnic groups. Asian women are just as likely to be in a romantic partnership as white women. Black women are marginally less likely to be in any relationship than white women (0.78 vs 0.84). White and Asian women have higher levels of SES (parental SES and educational attainment) than Hispanic and black women. However, the Wave I interviewers rated black women least favorably on attractiveness and grooming. Asian women, like Asian men, are the group with the lowest prevalence of romantic involvement during adolescence. In sum, while patterns of economic resources and adolescent romantic involvement for Asian men and women are similar, a sizeable gap exists between Asian men and Asian women with respect to those who did not report a current romantic involvement at Wave IV (35% vs 18%). Note that this pattern is the opposite of what the cultural framework explanation would predict.

To better highlight racial and ethnic differences in romantic involvement, and to identify structural or cultural factors that promote and impede involvement, we turn to multivariate analyses. Tables 3 and 4 present the odds ratios from logistic regression models predicting the likelihood of current relationship involvement in this sample of youth on the different-sex mate market. Following previous research on relationship involvement, we estimate separate models for men and women. We present three sets of models for each to understand whether taking account of structural factors such as economic resources or physical characteristics, or cultural factors such as nativity status or ethnic sub-group may explain the racial and ethnic gaps in romantic involvement found in the bivariate statistics.

Table 3.

Odds Ratios of Current involvement at Wave 4, MALE (N=5,840)

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Race/Ethnicity
Black	0.738	0.763	0.738	0.764	0.738	0.769
Hispanic	0.895	0.967
Asian	0.474 ***	0.586 *
Foreign Born		0.738				0.696
Foreign Born/Race/Ethnicity
US Born Hispanic			0.874	0.883
Foreign Born Hispanic			0.966	0.954
US Born Asian			0.474 ***	0.557 **
Foreign Born Asian			0.475 *	0.463 *
Hispanic Ethnic Group
Mexican					0.883	0.939
Cuban					1.164	1.254
Puerto Rican					1.232	1.343
Central/South American					1.272	1.622
Other Hispanic					0.461 *	0.500 *
Asian Ethnic Group
Chinese					0.299 ***	0.368 ***
Filipino					0.658	0.890
Other Asian					0.443 **	0.540 *
Wave 1 Variables
Parental SES		0.944 **		0.945 **		0.942 **
Living with both biological parents		1.208		1.207		1.233 *
IR attractiveness		1.150 ***		1.150 ***		1.150 ***
Romantic relationship		1.658 ***		1.665 ***		1.644 ***
Wave 4 Variables
Age (in years)		1.070 *		1.067 *		1.071 *
Height (in inches)		1.035 *		1.035 *		1.036 *
College graduate prior to age 25		1.145		1.140		1.160
Intercept	3.97 ***	0.011 **	3.97 ***	0.012 **	3.97 ***	0.010 **
F	5.77	8.47	3.81	7.94	5.03	6.94

^*p<.05;

^**p<.01;

^***p<.001

Note: All models are of current involvement with a different-sex partner among respondents who identify as 100% straight and report no involvement with a same-sex partner.

Table 4.

Odds Ratios of Current Involvement at Wave 4, FEMALE (N=5,715)

Variable	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Race/Ethnicity
Black	0.656 ***	0.706 **	0.656 ***	0.703 **	0.656 ***	0.704 **
Hispanic	0.875	1.022
Asian	0.850	1.153
Foreign Born		0.767
Foreign Born/Race/Ethnicity
US Born Hispanic			0.931	1.036
Foreign Born Hispanic			0.739	0.762
US Born Asian			1.054	1.131
Foreign Born Asian			0.735	0.897
Hispanic Ethnic Group
Mexican					0.957	1.124
Cuban					0.726	0.766
Puerto Rican					0.616	0.680
Central/South American					1.160	1.593
Other Hispanic					0.659	0.780
Asian Ethnic Group
Chinese					0.946	1.427
Filipino					1.200	1.660
Other Asian					0.683	0.959
Wave 1 Variables
Parental SES		0.975		0.975		0.974
Living with both biological parents		1.019		1.018		1.008
IR attractiveness		1.077 **		1.077 **		1.077 **
Romantic relationship		1.936 ***		1.936 ***		1.954 ***
Wave 4 Variables
Age (in years)		1.008		1.009		1.006
Height (in inches)		0.994		0.994		0.995
College graduate prior to age 25		1.112		1.112		1.118
Intercept	5.34 ***	2.02	5.338 ***	2.02	5.34 ***	1.95
F	5.11	6.02	3.47	5.48	2.13	3.94

^*p<.05;

^**p<.01;

^***p<.001

Note: All models are of current involvement with a different-sex partner among respondents who identify as 100% straight and report no involvement with a same-sex partner.

Table 3 presents the results for men. The zero-order model (Model 1) shows the unadjusted racial and ethnic gaps in current relationship involvement. Asian men, but not black or Hispanic men, exhibit significantly lower odds of involvement than white men. Specifically, they have roughly half the odds of current involvement as white men. The full model (Model 2) adds the independent variables. Results suggest that involvement increases with age, but not educational attainment, parental SES or nativity status. All of the variables targeting opportunities for involvement for men (i.e., attractiveness at Wave I, adolescent romantic involvement, and height) have effects in the expected direction. (We examined the addition of height squared to capture any nonlinearities but it failed to reach statistical significance). Importantly, the inclusion of these variables in the full model diminishes differences by race and ethnicity, but the Asian indicator remains significant, suggesting that the relative socioeconomic advantage of Asian American men does not translate into higher levels of partnering than any other race/ethnic group. Even in the descriptive tabulations, we can see that the higher SES attainment of Asian men do not transform into more favorable outcomes on the mate market – hence our results do not support the structural framework for accounting for the difference between Asian and white males.

To examine whether cultural preferences in immigrant families may account for race and ethnic gaps in partnering, we distinguish Asians and Hispanics by their nativity status. Model 3 depicts the zero-order while Model 4 represents the full model. (Given the small numbers of foreign-born white and black men in the sample, we do not distinguish these groups by nativity. We ran a parallel model in which they were omitted and the results are the same). The results suggest that Asian men regardless of nativity status have substantially lower odds of forming a romantic partnership at Wave IV than do whites. No differences by nativity status were found among Hispanic men. These results do not support the cultural framework model.

To further explore whether perhaps cultural variations may be driving this gap in relationship involvement, we divide Asian and Hispanic men into ethnic sub-groups (Models 5 and 6). Based on sample size limitations, we break Asians into Chinese, Vietnamese, and Other Asian; and Hispanics into Mexican, Cuban, Puerto Rican, Central/South American or Other Hispanic. The zero-order model (Model 5) suggests that men reporting Chinese or “Other Asian” ethnicity have substantially lower odds or romantic involvement than whites. On the other hand, Filipino men are or no more or less likely to form a romantic partnership than white men. We suspect that this may be because Filipinos are often seen as the “Hispanics of Asians.” Previous research suggests that they may feel great affinity with Latinos than Asians, and are not seen as “model minorities.” Thus, it is possible that Filipinos are less likely than other Asians to be seen as geeky and undesirable dates (Ocampo, 2013a, 2013b). In addition, among Hispanics, the group “Other Hispanic” displays significantly lower odds of relationship involvement than whites. The addition of the full set of independent variables attenuates the association but does not eliminate it. We also considered another aspect often attributed to culture—whether Asian men may be excluding themselves from forming romantic relationships due to enrollment in higher education. In models (not shown) we interacted current enrollment in college measured at the fourth wave, with race/ethnicity and found no significant associations. In sum, our results lend little support for a cultural explanation of race and ethnic gaps in partnering.

We estimated parallel models for women (Table 4). Results from the zero-order model for women (Model 1) reveal that black women have significantly lower odds of current involvement than white women. Two of the factors that differentiate the involvement of men also differentiate the involvement of women—interviewer-rated attractiveness and prior romantic relationship. Perhaps reflecting women’s earlier involvement in relationships, age does not have a significant nor positive effect. Height fails to significantly affect the involvement of women. The full model (Model 2) reveals that net of resources and characteristics, black women have lower odds of romantic partnering than white women. Hispanic and Asian women show no difference in their odds of current involvement at Wave IV compared to white women. In contrast to the findings concerning men, no differences in involvement for Asian or Hispanic women were found in models examining nativity status (Models 3 and 4), ethnic sub-group (Models 5 and 6) or current educational enrollment (not shown).

DISCUSSION

This study reassesses racial and ethnic patterns of current romantic involvement, regardless of co-residential union status. We use a broader definition of romantic relationships and a more exhaustive set of race/ethnic groups (i.e., whites, blacks, Hispanics, and Asians) in our examination of the patterns and correlates of any romantic relationship involvement in adulthood in order to understand why some young adults remain unpartnered. We draw from structural, cultural, and critical race frameworks to develop expectations about racial and ethnic gaps in romantic involvement, paying close attention to the distinctiveness of Asians Americans.

Our descriptive statistics reveal that patterns of involvement differed markedly for men and women. Asian men were much more likely than their same-sex counterparts from other race/ethnic groups to be unpartnered. Roughly one out of every five Hispanic and white men failed to report a current sexual and/or romantic partner; only about a quarter of black men and a third of Asian men were similarly classified. White, black, and Hispanic women generally resembled their male counterparts in their levels of involvement, but Asian women were half as likely as Asian men to be unpartnered (i.e., 18% versus 35%).

In zero-order models run separately for men and women, black women and Asian men consistently exhibited a significantly lower likelihood of current involvement than their white counterparts. In full models that included economic factors, physical attributes and prior adolescent romantic relationships, the patterns remained. Wave I interviewer-rated attractiveness and romantic relationship involvement additionally reduced involvement for men and women, underscoring the importance of experience and opportunity. We found no evidence that socioeconomic resources or physical characteristics were driving the lower levels of involvement among Asian men. Instead, our findings are consistent with the notion that Asian American men are at the bottom of the racial hierarchy when it comes to the different-sex dating market.

With our data, we are unable to completely determine the mechanism through which Asian men are excluded from the mate market at this stage of the life course. While we cannot definitively say that Asian American men have not removed themselves from the mate market, we find little support for that argument. The significantly lower prevalence of involvement among Asian men, regardless of nativity status and ethnic group, challenges the notion that cultural norms and values may account for race and ethnic patterns in current involvement. Filipino men are an exception to this pattern, suggesting that they are more like Hispanics than all other Asian groups in terms of romantic/sexual involvement (Ocampo, 2013). Further evidence against the cultural argument is found in the stark differences in the likelihood of being unpartnered between Asian men and women. If “Asian cultural values” account for differences in romantic partnership formation, we would have found similar patterns for Asian women as Asian men (or alternatively, lower levels of involvement among Asian American women than Asian American men). In fact, the opposite was true. Finally, if culturally-specific values and norms were driving romantic involvement, these differences would be more apparent among the foreign-born population than among the second and later generations.

Some of this gap could be due to the large gender differences in interracial involvement among Asians. Prior studies reveal, among other things, that Asian women are more likely to form co-residential relationships with someone of a different race than their Asian male counterparts and that black men are more likely than black women to interracially co-reside (Qian and Lichter 2007). Similarly, in analyses not shown, we find that given a relationship, Asian men in our sample were less likely than Asian women to be in a relationship with a different-race partner (i.e., 33% versus 10%). However, research by Fong and Yung (1995) suggests that both Asian men and women express a similar desire to marry outside of their race. The differences we identified between white and Asian men were consistent with notions of a racial hierarchy in dating preferences, as well as observations by Asian American film scholars and media observers (Adachi 2008).

Media observers continue to note that Hollywood movies and advertisements promote Asian American men as geeky and awkward romantic partners. Asian American males usually serve as exemplar in their awkwardness around women. The character Long Duk Dong from Sixteen Candles (1984), a film by John Hughes, has been widely viewed as the quintessential Asian American geek and is well known to Asian Americans males who were adolescents in the 1980s. As one literary source stated, “Every single Asian dude who went to high school or junior high during the era of John Hughes movies was called ‘Donger’” (Wong and Nakamura quoted in MacAdam, 2008). Film scholars and Asian American commentators have also lamented the fact that Asian American men rarely play romantic leads in films. Even when Asian American men are present as the lead male (usually in martial arts or action films), they rarely have the opportunity to demonstrate any physical displays of affection. For example in Romeo Must Die, a film based on Romeo and Juliet starring Chinese actor Jet Li and Aaliyah (an African American actress), the two leading characters never kiss despite the romantic link between them. A kiss was tested with a focus group, but they were uncomfortable with it, so in the cinematic release of this film, the two characters (again, based on Romeo and Juliet) hug instead.

The analyses presented in this study are limited in a few respects. First, we do not measure mate availability. While structural frameworks do stress the importance of individual characteristics (i.e., economic resources, physical attractiveness) in forming a relationship, they also focus on the demographic availability of potential mates. Research suggests that black women’s likelihood of being married decreases as black men’s rates of interracial marriage in their metropolitan area increase (Crowder & Tolnay 2000). Indeed, prior research finds that higher sex ratios (i.e., greater numbers of men relative to women) are associated with lower rates of marriage (Angrist 1990; South and Lloyd 1992; Fossett and Kiecolt 1991; Lichter et al 1992; Lichter, LeClere and McLaughlin 1991). However, prior studies concerning mate availability find that sex ratio imbalances explain only a small part of the marriage gap between black and white women (e.g., Lichter et al. 1992; Mare and Winship 1991), and that current relationship involvement among young adults is largely unaffected by partner availability (Warner, Manning, Giordano, and Longmore 2011). In the current study, we are unable to address the influence of sex ratios on racial and ethnic patterns of current involvement because the contextual data corresponds only to the period of interviews. More specifically, the contextual data in the fourth wave include only the number of women and men for the ages 18 to 24, not the full age range that might compose a potential mate pool.

A second limitation concerns our measurement of cultural norms. We do not have measures of parental attitudes toward the formation of romantic partnerships, which might provide us with a more direct measure of family influences on young adult behaviors. Still, it is unlikely that Asian American parents would be more restrictive with their sons than daughters (Dasgupta 1998; Espirtiu 2001; Talbani and Hasanali 2000). However, we did explore the role of several variables measured at Wave I in mediating the gap between Asian and white men in our preliminary models (not shown here). Specifically, we measured parents’ disappointment if their child were not to graduate from college based on the parent questionnaire (and alternatively respondent reports). Alternatively, we considered the effects of respondents’ own educational aspirations and expectations (combining the two items on a scale). We also included measures of verbal aptitude and grade point average. Finally, we measured attitudes among young adults related to union formation during the third wave (e.g., the importance of being married someday and the importance of partner race for a successful relationship). None of the variables we considered reduced the magnitude of the gap substantially or reduced its level of significance. As Add Health does not ask respondents about their willingness to form relationships with partners from specific racial groups, we are not able to empirically test the racial hierarchy explanation and suggest that future population-based surveys incorporate these measures.

Nonetheless, our results do suggest that a racial hierarchy in romantic partner preferences may hinder Asian men from entering into romantic partnerships during young adulthood. Published U.S. Census statistics on marriage suggest that Asian American men eventually marry; among Asian American men ages 40 to 50 years, just 12% are reported as never married compared with 16% of white men. However, these figures include Asian men who dated and married outside of the US – hence, they would not have suffered under the same racial hierarchy as they might have had they dated in the US. We do not know whether this younger cohort of Asian American men will eventually marry at similarly high rates or whether the patterns of exclusion found in the dating market here will continue to the marriage market. Data from the US Census also includes individuals that married outside of the US, so even if Asian American men are marginalized in the US dating market, that would not affect the odds of marriage among Asians outside of the US. It is also possible that Asian American men are disadvantaged in the dating market, but eventually marry due to their higher SES attainment.

Our study is an important first step into understanding the marginalization of some groups in the dating market, but additional studies are needed that offer a more comprehensive view of romantic involvement ranging from where individuals search for and meet potential partners to how the romantic relationship progresses at this stage of the life course. Prior studies have suggested that social context shapes the networks from which individuals chose or are chosen as mates (Kalmijn and Flap 2001; Laumann, Gagnon, Michael and Michaels 1994). Where individuals meet romantic partners (i.e., school, work, Internet) influences not only the progression of relationships but the perceived support for them (Sassler and Miller 2014). Examining factors such as these may help to explain some of the gaps we find in romantic involvement not only for Asian men but for black women. Future studies should also examine whether the race/ethnic and gendered exclusion we find extend to the same-sex population as suggested by prior research using select samples (Phua and Kaufman 2003; Tsunokai, McGrath and Kavanagh 2014).