Young Adult Relationship Values at the Intersection of Gender and Sexuality

Abstract

Recent decades have brought significant social changes in the industrialized West that may influence young adults’ attitudes about intimate relationships, including changes in gender expectations and behaviors and changes in sexual attitudes and practices. We used data from the National Longitudinal Survey of Adolescent Health (N=14,121) to compare men to women, and sexual minorities to heterosexuals, on ratings of the importance of love, faithfulness, commitment, financial security, and racial homogamy for successful relationships. We found that nearly all young adults adhere to dominant relationship values inherent in the romantic love ideology; however, we found modest but significant differences by gender and sexual identity in relationship values. Significant interactions demonstrated that gender and sexual identity intersect to uniquely influence relationship views.

One of the most important life-course events of young adulthood is the formation of intimate relationships. The last several decades have brought significant social changes in the industrialized West that may influence young adults’ attitudes about intimate relationships, including changes in gender ideologies and sexual attitudes and practices. Women’s identities are not as narrowly defined by romantic ties and family roles as in the past (Gerson, 2001). Attitudes about premarital sex, cohabitation and homosexuality have liberalized, expanding sexual and relationship possibilities (Johnston, Bachman, & O’Malley, 2007; Loftus, 2001; Thornton & Young-DeMarco, 2001). As a reflection of these changes, cultural, political and legal battles about same-sex marriage have intensified, and differential access to marriage among sexual minorities (gays, lesbians, and bisexuals) may shape young adults’ views of relationships.

With changes in the social context and heightened public interest in and debate about same-sex relationships, we take stock of the relationship views that a new generation of young adults holds. Do young people’s relationship values still differ by gender, or have recent trends toward gender equality and the erosion of traditional ideologies erased differences? With more opportunities to pursue a range of relationships, the once transgressive relationships of sexual minorities are less transgressive today as evidenced by research interest in same-sex relationships (Allen & Demo, 1995). Are the relationship values of heterosexuals and sexual minorities similar or different? Do gender and sexual identity intersect to shape relationship values?

Using data from the National Longitudinal Study of Adolescent Health (Add Health), we assess relationship values by gender and sexual identity among 18 – 28 year olds. In what follows, we first review past research on relationship values as well as relevant gender and relationship theories. Based on this work, we develop hypotheses about similarities and differences in relationship values by gender and sexual identity. Next, we describe our data source, measurement techniques, and method of analysis. Finally, we present our results and discuss the implications of our findings.

Past Research on Relationship Views

Marriage and Relationship Views of Youth

American youth’s relationship values have remained fairly consistent over several decades. From the mid-1970s to the late 1990s, 80% of female high school seniors and 70–75% of male seniors reported that they expected to marry eventually, and similar percentages rated having a good marriage and family life as “extremely important”. Somewhat lower percentages (62–69% of young women and 55–57% of young men) reported that it was “very likely” they would remain married to the same person for life (Thornton & Young-DeMarco, 2001). More recent data are consistent: 80–90% of youth expect to marry, 70–80% say marriage and family life is “extremely important,” and 60% say lifelong marriage is “very likely” (Johnston et al., 2007). Gender differences are modest yet consistent: young women express higher expectations to marry, greater importance of marriage and family life, and higher expectations for lifelong marriage (see too Harris & Lee 2006). This suggests that young women adhere to traditional relationship values more than young men.

Research on college students’ mate preferences has found that men historically ranked love and attraction higher than women as a desired mate characteristic, but the gender difference recently declined, and women and men now rate love and attraction as the most important quality. At the same time, another characteristic—ambitiousness/industriousness—dropped in value for women (Buss, Shackelford, Kirkpatrick, & Larsen, 2001). These shifts may reflect women’s growing economic independence and the declining importance of a male partner’s earning ability. Yet, women still ranked ambitiousness more highly than men, suggesting a slightly higher value on characteristics that foretell financial security. Beliefs about the benefits of marriage have also been consistent across time: both genders rated love and affection most important, although women assigned greater value to this than men (Barich & Bielby, 1996).

In her research on how middle-class adults talk about love, Swidler (2001) concluded that most Americans remain attached to the romantic love myth, even in the face of high divorce rates and a growing “realist” discourse about love as difficult and contingent. Swidler identified four main features of the romantic love myth – love as a clear and decisive choice (“love at first sight”), love as unique and exclusive (“one true love”), love as surmounting social obstacles (“love conquers all”), and love as enduring (“happily ever after”) – and argued that the structural realities of the institution of marriage account for the ongoing cultural power of the romantic love myth (Swidler, 2001, pp. 117–124).

Regarding status traits of potential partners, past research has shown that among adults, women emphasize intellect, employment, and financial status in desired companion characteristics whereas men emphasize physical attributes (Davis, 1990). Although with women’s increase in economic independence, their partners’ material status may be less essential for the sustainability of the relationship (Oppenheimer, 1988). Although not about mate preferences, other research has shown that female youth are more accepting than male youth of interracial contact through friendships, neighborhoods, and work (e.g., Kirkpatrick & Marini, 1998). This tolerance likely extends to romantic partners as well.

Relationship Beliefs and Practices of Sexual Minorities

Several studies have found that partners in same-sex relationships are less socially similar on average than heterosexual partners on dimensions of race/ethnicity, SES, and age (Jepsen & Jepsen, 2002; Kurdek & Schmitt, 1987; Rosenfeld, 2007). These studies provide indirect evidence that sexual minorities have more open attitudes about forming relationships with socially dissimilar partners. It could also be the case, however, that sexual minorities have a smaller pool of eligible partners, forcing them to relax partner requirements (Ellingson, Laumann, Paik, & Mahay, 2004). In addition, Black and colleagues (2000) found that lesbians earn more than other women, so perhaps financial security in relationships is less of a concern for them. In short, concerns about status issues such as race and money may be less important for sexual minorities than for straight young adults.

A second important finding is that same-sex relationships have shorter average duration than heterosexual relationships (Kurdek, 2004), and female same-sex relationships may be shorter than male same-sex relationships (Carpenter & Gates, 2008). Differences in relationship dissolution rates for sexual minorities may reflect the lower social and legal support for same-sex couples and/or lower value placed on long-term commitment by sexual minorities. Observed differences in relationship stability may also influence the aspirations and expectations of sexual minorities, leading them to view lifelong commitment as less desirable or achievable.

Despite their higher dissolution rates, female couples engage in wedding-like public commitment rituals more often then male couples (Hull, 2006), and in jurisdictions where marriage-like legal partnership status is available, female couples enter such partnerships at higher rates than male couples (Carpenter & Gates, 2008). These findings challenge the idea that lesbians value commitment less; they may value commitment as much as others, but for various reasons, find relationship permanence difficult to achieve.

A third set of findings concerns sexual fidelity. Several studies indicate that male same-sex couples practice sexual monogamy less than heterosexual or lesbian couples (Adam, 2006; Kurdek, 1991), and one study found that male same-sex couples rate fidelity as less important than others (Kurdek, 1991). Some gay men may decouple faithfulness from sexual monogamy, with faithfulness referring to the emotional primacy of a relationship rather than its sexual exclusivity (Adam, 2006). These studies suggest that both sexual identity and gender shape commitment to the ideal of sexual monogamy.

Little research exists on relationship practices and views of sexual minority youth, but we know that they are less likely than heterosexuals to experience any romantic relationships during adolescence (Diamond & Dubé, 2002). Relationship experiences of sexual minority youth may vary by gender, with male same-sex relationships more likely to begin as sexual liaisons with less closeness, attachment and support, and lesbian relationships more likely to start as friendships (Diamond, 2003). A small-scale study of gay male youth found commitment to the sexual script of romantic love and desire for relationships, but also a pattern of repeated casual sexual encounters, and openness to the idea of nonmonogamous relationships (Mutchler, 2000).

Researchers have examined other factors associated with youth’s marriage and relationship views and behaviors. Minorities, especially Blacks, are less likely to marry (Goldstein & Kenney, 2001) and less likely to date during adolescence, especially in serious romantic relationships that increase marital expectations (Crissey, 2005). Other research indicates that financial security is an especially important marital pre-requisite for poor women (Edin & Kefalas, 2005). As evidenced by the religious underpinnings of the covenant marriage movement, more religious individuals place a higher premium on commitment in relationships (Nock, Sanchez, & Wright, 2008). Finally, the structure and stability of families of origin are related to offspring relationship beliefs (Sprecher, Cate, & Levin, 1998). In an analysis of Add Health data, Harris and Lee (2006) found that race/ethnicity, socioeconomic status, family background, religiosity, and previous family formation experience are associated with one or more measures of marriage and relationship views (they do not explore sexual identity). Although not central to our study, we include these factors as controls. We also include measures for political orientation and involvement in an effort to insure that any association between sexual identity and relationship values is not simply a result of the more general liberal orientation of sexual minorities. We expect gender and sexual identity to shape relationship views net of these other important influences.

Theoretical Perspectives on Gender, Sexuality and Relationship Attitudes

In this section we review leading theories of gender and sexuality in order to develop hypotheses and a theoretical framework in which to understand our findings about young adult relationship values. Our intention is not to test the relative merits of these theories, but rather to combine aspects of each perspective to achieve a deeper understanding of our findings.

Gender Theories

Prominent approaches to studying gender provide theoretical frameworks for understanding how dominant cultural norms such as feminized love produce, and are reproduced by, gender-typed attitudes and practices. The gender socialization framework includes psychoanalytic (Chodorow, 1978), social learning (Mischel, 1966) and cognitive theories of gender development (Bem, 1981; Maccoby & Jacklin, 1974). All of these suggest gender identity formed through early socialization is a fairly stable characteristic of individual personalities in adulthood. Current perspectives on gender view the socialization framework as too limited because it treats gender as a stable individual characteristic and neglects issues of power and structure (Connell, 1987; Kimmel, 2000). Two theoretical alternatives are the gender performativity and the gender-as-institution approaches. Gender performativity theories regard gender as emergent in social interaction, and therefore variable and dependent on situations. This “doing gender” approach (West & Zimmerman, 1987) suggests individuals adapt their gender presentation to the demands of specific interactions, and are aware of the social costs of a failed performance, that is, departing from dominant gender norms. This suggests that part of women’s gender performance will be the expression of interest and skill in intimate relationships and adherence to values celebrating commitment, fidelity, and love’s power to surmount obstacles, what Swidler (2001) called the romantic love myth. Although the myth influences all cultural actors, women are expected to enact it with particular enthusiasm and expertise.

Theories of gender as a social institution broaden perspectives on gender by highlighting issues of inequality, power and constraint, ideology, and the gendered nature of social institutions such as marriage, family, and workplace (Connell, 1987; Martin, 2004). This places socialization and gender performances in a broader context of inequalities that are reinforced by institutional arrangements (e.g., domestic division of labor, organization of paid work) and legitimating ideologies (e.g., essentialist accounts of male-female difference). The gendered beliefs, practices, and relationship values of individuals, then, are not solely the product of socialization or the exigencies of situated interactions but also of institutional pressures. These theories speak to expected gender differences in the relationship values of contemporary young adults. Despite different angles of vision, all of the theories suggest that women will be more invested in relationship issues and adherent to hegemonic relationship norms than men.

Sexual Minorities as Relationship Innovators

Three middle-range theories of sexual identity and relationships cast sexual minorities in the role of relationship innovators. First, Giddens (1992) argued that modern societies are shifting from a relationship model based on romantic love to one of confluent love and the pure relationship. Romantic love entails lifelong commitment and “a psychic connection, a meeting of souls” which completes individuals and produces a shared life trajectory (Giddens, 1992, p. 45). In confluent love, partners aspire to the pure relationship, “where a social relation is entered into for its own sake …and which is continued only in so far as it is thought by both parties to deliver enough satisfactions for each individual” (Giddens, 1992, p. 58). The rise of confluent love undermines ideas of commitment and fidelity; in pure relationships, everything is open to negotiation and no one is bound to the relationship out of duty. Giddens argued that same-sex relationships represent the vanguard of the shift to confluent love (p. 135).

Similarly, Cherlin (2004) extended his idea of incomplete institutionalization to same-sex relationships. He originally argued that remarriages are less stable than first marriages because they are incompletely institutionalized: partners face ambiguity regarding their roles and responsibilities toward each other and stepchildren (Cherlin, 1978). Extending the idea to same-sex partnerships, he noted: “Lesbian and gay couples who choose to marry must actively construct a marital world with almost no institutional support” (Cherlin, 2004, p. 851).

Finally, according to Rosenfeld’s (2007) age of independence theory, the latter part of the twentieth century saw the rise of an independent life stage, a period between leaving the parents’ home and forming one’s own family, in which young adults live independently and are free to pursue relationships that do not conform to parents’ expectations. As such, parents’ influence over mate selection has declined dramatically, as reflected in increases in interracial and same-sex relationships (Rosenfeld 2007, Figure 4.1). We extrapolate from Rosenfeld’s theory to predict that sexual minorities will be more open to relationships that violate dominant social norms, such as interracial relationships, since they have already violated the norm of heterosexuality and achieved considerable independence from their parents’ influence.

All of these theories point in the same direction—that sexual minorities are less tethered to traditional relationship beliefs and practices. Incomplete attachment to dominant norms—the romantic love ideology for Giddens and the institution of marriage for Cherlin—accounts for sexual minorities’ greater degree of relationship innovation. For Rosenfeld, the new independent life stage frees all young adults, but especially sexual minorities, to form relationships that deviate from the expectations of their families and communities. Specifically, heterosexuals should endorse traditional relationship values, especially those for commitment and fidelity, more strongly than sexual minorities. Further, sexual minorities should be less concerned with the race and SES of partners.

Intersectionality

The concept of intersectionality was originally developed by feminist and critical race theorists to call attention to the ways in which identity categories intersect to produce distinctive social experiences that are not reducible to their component parts (e.g., Davis, 1981; Glenn, 1985). The concept implies, for example, that African American women’s experience is not merely the sum of their experience as Blacks and as women, but rather that the experience of being Black has a different meaning for men and women, and the experience of being a woman is different for African Americans and those of other racial/ethnic groups.

Several of the ideas discussed above can be extended to consider how the intersection of two identity categories shapes relationship views. For example, the gender-as-institution theory asserts that gendered beliefs and practices are the product of institutional pressures. This perspective suggests that marriage, as the primary institution that governs intimate relationships, reinforces gender differences which reflect the hegemonic belief that roles of partners should be complementary or different. Because legal access to marriage is largely unavailable to same-sex couples, most sexual minorities are not governed by these same institutional forces. Therefore, their relationship views may not be as influenced by the hegemonic gender ideologies of institutions like marriage. Thus, the intersection of gender and sexual identity may produce unique relationship views; one’s gender may not shape relationship values for lesbian and gay young adults in the same ways that it does for straight young adults.

Much of the empirical work on intersectionality emphasizes its complexity, leading researchers to favor qualitative methods (McCall, 2005). But intersectionality is also amenable to quantitative inquiry with the use of interaction terms or subgroup comparisons to test whether the effects of holding one identity are conditioned by other identities. We use such methods to examine the intersection of gender and sexual identity with respect to relationship values.

Hypotheses

Based on past research and theories reviewed above, we offer the following general hypotheses regarding the association of gender and sexual identity with young adults’ relationship values:

• H1Gender: Young adult women will place more importance on traditional relationship values such as love, faithfulness, and lifelong commitment than their male counterparts.
• H2Sexual identity: Sexual minorities will place less importance on traditional relationship values—especially faithfulness and lifelong commitment—than heterosexuals, but they will express more support for nontraditional relationships such as interracial relationships.
• H3Intersectionality: Gender and sexual identity will intersect such that the gender differences found among heterosexuals will not always hold among sexual minorities.

METHOD

Data

We use data from the National Longitudinal Study of Adolescent Health. Adolescents in 80 high schools and 52 middle schools were sampled to yield a nationally representative survey of U.S. adolescents in grades 7–12 in 1994–1995 (Udry, 2003). In 1995, about 20,000 youth and their parents completed an in-home interview (Wave 1). About 14,700 also completed in-home interviews in 1996 (Wave 2), and 15,197 did so in 2001–2002 (Wave 3). These data are particularly well-suited to our research questions for several reasons. Unlike many other studies, Add Health includes a measure of sexual identity and a rich set of relationship values as described below. In addition, the sample is large enough to support multivariate analysis of sexual identity-by-gender groups—groups often too small to detect significant differences in analysis of the smaller samples common to most other studies.

We use measures from Waves 1 and 3 resulting in a sample of 15,197 before restrictions; most background factors are measured at Wave 1, gender, sexual identity, all relationship values and the remaining background factors are measured at Wave 3. We use data from all respondents who completed the Wave 3 interview and who had valid sexual identity, sample weight and survey design measures. Sample weights and design measures are missing (N=860; 5.65%) if the case was not in the original sampling frame but was added in the field, if the case was selected as part of a pair (twins, half-siblings) and both were not interviewed, or if the case did not have an indicator for whether it was part of a special oversample (Tabor, 2003). We also ran unadjusted models that included cases with missing sampling weights. Our main findings remained unaltered. Item non-response on sexual identity self-reports resulted in the loss of an additional 140 cases (1%), and asexual respondents—those who are not attracted to men or women—were dropped from the analysis (N=76; 0.5%). After these restrictions, our analysis sample is 14,121. Sample sizes for specific analyses vary slightly based on the degree of item non-response for each dependent variable (ranging from 10–18 cases).

Our sample is evenly split between men and women. The average age at Wave 3 is 22 (range 18–28). Sixty-eight percent of respondents are non-Hispanic White, 16% are non-Hispanic Black, 12% are Hispanic, and 4% are non-Hispanic Asian or Pacific Islander. Over 75% and 30% of respondents have parents who are high school and college graduates, respectively. Fifty-seven percent of respondents come from continuously married parent families.

Measures

Gender and Sexual Identity

Gender is self-reported by respondents as either man (0) or woman (1). We use a Wave 3 question that asks respondents to choose the description that best fits how they think about themselves in terms of their sexual identity. We code them into three groups: 1) straight – ‘100% heterosexual (straight)’ or ‘mostly heterosexual (straight), but somewhat attracted to people of your own sex’; 2) bisexual – ‘attracted to men and women equally’; and 3) gay/ lesbian – ‘100% homosexual (gay)’ or ‘mostly homosexual (gay), but somewhat attracted to people of the opposite sex.’

Because there are a number of ways to define sexual identity given self-reported identity and questions about sexual behavior and attraction, we ran models with several different specifications. In the first set of alternative specifications, we grouped the five response options in two different ways: 1) combining “mostly heterosexuals” and “mostly homosexuals” with “bisexuals”; and 2) combining “mostly heterosexuals” with “100% heterosexuals” but “mostly homosexuals” with “bisexuals.” We tried the latter specification because in cross-tabulations with our dependent variables, those who classified as “mostly heterosexual” responded in ways similar to the “100% heterosexuals,” and those who classified as “mostly homosexual” responded in ways similar to the “bisexuals.” In another alternative specification we defined sexual identity by substituting prior sexual behavior and attraction measures where self-reported sexual identity was missing. In a final specification, we defined sexual identity by including as sexual minorities those who identify as straight in Wave 3 but who had prior experience with or attraction to someone of the same sex. In models with all of these different specifications our results remained similar to what we report below. Ultimately, we define sexual identity to reflect what we believe to be the most reasonable combinations of categories given the question wording and so that sexual identity-by-gender groups are large enough to support multivariate analysis. Moreover, because we are interested in relationship views in the context of social and legal attention to same-sex relationships, sexual identity (not behavior) is the appropriate component of sexual orientation to assess (Savin-Williams & Ream, 2007).

Relationship Values

We use five measures of relationship values as dependent variables. The values are prefaced with the question: “How important do you think each of the following elements is for a successful marriage or serious committed relationship?” The elements are: being of the same race, having enough money, love, faithfulness, and making a lifelong commitment. The importance of racial homogamy and financial security are conceptualized as values about status features of relationships because they signal status in the broader society. The importance of love, faithfulness, and lifelong commitment are conceptualized as values about affective dimensions because they indicate feelings that are generally internal to individuals or between partners.

In their original scale, each measure ranges from 1 to 10 where 1 = not at all important and 10 = extremely important. We view these measures as ordinal, not interval because we cannot be sure that the distance between adjacent categories is equal across the scale (Long 1997). Some values on the scales have very few cases, especially for sexual minorities. In coding these measures, we seek to retain as much gradation as possible but still ensure that each category contains enough cases for meaningful analyses. Thus, we collapse the scales into smaller sets of categories as described below. We tested models using ordinary least squares regression with the measures in their original scales, and results are very similar.

For status value measures (racial homogamy and financial security), responses cluster at the two extremes and the middle of the distribution. We collapse the scale into four categories: 1 = not important; 2 = slightly important (2–3 on the original scale); 3 = somewhat important (4–6 on the original scale); and 4 = important to extremely important (7–10 on the original scale).

The affective value measures (love, faithfulness, and lifelong commitment) have highly skewed distributions; an overwhelming majority of respondents rate all three as extremely important (86, 89, and 76% respectively). Because we are interested in gender-by-sexual-identity specific associations, these scales must have enough cases in each category for each sexual-identity-by-gender group. For the love measure, we dichotomize the scale because we were not able to create more than two categories that contained enough cases to support regression analyses. Those who indicate that love is extremely important (10) are coded 1; all others are coded zero. For faithfulness and lifelong commitment, we retain three categories indicating: 1 = not important to important (1–7 on the original scale); 2 = very important (8–9 on the original scale); and 3 = extremely important (10 on the original scale). Even the lowest category on these measures includes those who rate lifelong commitment as important; thus, the categories should be interpreted as gradations in the top of the original distribution.

Background Factors

We use a number of controls: race/ethnicity, adolescent family structure, and family socio-economic status all measured at time 1, and religiosity, political orientation and involvement, age, and whether the respondent has ever married, cohabited, or had a child, all measured at time 3. For the sake of brevity, we do not detail the measurement of these variables here (available upon request), although they are shown in Table 1.

Table 1.

Descriptive Statistics for Independent Variables (weighted and design adjusted)

Panel A: Key Independent Variables
	Percent		Percent	Unwgt N
Sexual identity		Sexual identity × gender
Straight (ref)	96.95	Straight women (ref)	47.20	7162
Gay/lesbian	1.50	Straight men	49.75	6512
Bisexual	1.55	Lesbians	0.58	84
		Gay men	0.93	134
Gender		Bisexual women	1.26	187
Men (ref)	50.97	Bisexual men	0.30	42
Women	49.03
Panel B: Background Variables
	Percent		Percent
Race		Family of origin
Non-Hispanic White (ref)	67.55	2-parent, bio/adoptive (ref)	57.13
Non-Hispanic Black	15.54	Other family structure	42.87
Hispanic	11.64
Non-Hispanic Asian	3.65	Parent’s Education
Non-Hispanic other	1.62	< HS grad	24.79
		HS grad/some college (ref)	40.05
		College grad+	32.45
Political orientation		Missing	2.71
Liberal	17.51
Moderate (ref)	52.30	Family formation history
Conservative	20.35	Married, intact	17.00
Missing	9.84	Married, divorce, no remarry	1.62
		Ever cohabit	41.04
Registered to vote	71.39	Ever parent	20.09
Religion		Ever on public assistance	8.15
Conservative Protestant	13.24
Moderate Protestant (ref)	9.46		Mean
Liberal Protestant	4.54	Religiosity Index (range)	4.34 (0 – 9)
Black Protestant	7.97
Catholic	23.64	Age (range)	21.81 (18 – 26)
Jewish	0.76
Other religion	2.64
No religion	22.00	Average log of family income	3.57
Affiliation missing	15.74	Missing Family Income	21.90%

Note: N = 14,121; Population Size = 21,706,868. Reference category denoted by (ref).

Method of Analysis

Because our dependent variables are ordinal, we use ordered logistic regression for four of the five outcome measures. We use logistic regression for the importance of love which we treat as dichotomous as noted above. In Table 2 we show the distributions of relationship values overall, and by gender and sexual identity. In Tables 3 and 4, we show associations between relationship values, gender, sexual identity, and controls in logistic and ordered logistic regression models. Model 1 shows associations between each relationship value and gender, sexual identity, and control measures. We then test gender-by-sexual-identity interactions (not shown). If these interactions are significant, Model 2 explicitly tests differences between each sexual-identity-by-gender subgroup using post-hoc Wald tests. Figure 1 illustrates key interaction findings. We adjust for the Add Health sampling design using the svy commands in STATA 9.2. We estimate associations but cannot test causal effects because sexual identity and our outcomes are both measured at Wave 3.

Table 2.

Weighted Percentage Distributions of Relationship Values by Gender & Sexual Identity

		Importance of same race
	Overall total %	Men			Women
		Straight	Gay	Bisexual	Straight	Lesbian	Bisexual
Not important	44.47	41.89	57.00	56.75	46.20	49.72	67.18
Slightly important	12.64	12.70	12.14	25.24	12.44	18.12	12.29
Somewhat important	19.20	20.30	14.84	0.33	18.40	25.93	10.37
Impt to very impt	23.69	25.11	16.02	17.68	22.95	6.23	10.16
		Importance of having enough money
	Overall total %	Men			Women
		Straight	Gay	Bisexual	Straight	Lesbian	Bisexual
Not important	9.40	11.16	10.75	10.41	7.21	8.65	19.87
Slightly important	7.92	7.39	10.86	20.41	8.08	24.48	9.94
Somewhat important	32.71	31.38	46.14	34.35	33.79	28.37	36.42
Impt to very impt	49.98	50.05	32.25	34.83	50.92	38.50	33.77
		Importance of lifelong commitment
	Overall total %	Men			Women
		Straight	Gay	Bisexual	Straight	Lesbian	Bisexual
Not impt to impt	10.82	14.00	17.59	31.92	6.65	25.48	24.59
Very important	12.80	14.57	18.81	11.67	10.57	26.89	15.31
Extremely important	76.39	71.43	63.60	56.41	82.79	47.63	60.10
		Importance of love
	Overall total %	Men			Women
		Straight	Gay	Bisexual	Straight	Lesbian	Bisexual
Not extremely impt	13.91	18.41	17.86	20.98	8.85	26.91	14.26
Extremely important	86.09	81.59	82.14	79.02	91.15	73.09	85.74
		Importance of faithfulness
	Overall total %	Men			Women
		Straight	Gay	Bisexual	Straight	Lesbian	Bisexual
Not impt to impt	3.37	4.71	9.06	16.74	1.76	3.55	3.03
Very important	7.80	10.26	10.41	9.39	4.89	16.66	13.31
Extremely important	88.83	85.02	80.53	73.87	93.36	79.78	83.66

Note: distributions are adjusted for complex survey design using STATA 9.2 svy commands

Table 3.

Ordered Logistic Models of Importance of Status Values

	Same race		Having enough money
	B	SE B	B	SE B
Woman	−0.12 (0.89)*	0.05	0.17 (1.18)**	0.05
Gay/lesbian	−0.36 (0.70)*	0.17	−0.55 (0.58)**	0.14
Bisexual	−0.58 (0.56)**	0.18	−0.59 (0.55)**	0.18
Non-Hispanic Black	−0.14 (0.87)	0.10	1.03 (2.80)**	0.09
Hispanic	−0.38 (0.68)**	0.09	0.52 (1.68)**	0.10
Non-Hispanic Asian	−0.22 (0.80)	0.14	0.57 (1.77)**	0.14
Non-Hispanic other race	−0.30 (0.74)	0.17	0.61 (1.84)**	0.19
Age (wave 3)	0.08 (1.08)**	0.02	0.04 (1.04)**	0.01
Alternative family structure	−0.21 (0.81)**	0.05	−0.05 (0.96)	0.06
Parent ed, <HS grad	0.16 (1.18)**	0.06	0.06 (1.06)	0.07
Parent ed, college grad	−0.05 (0.95)	0.05	−0.12 (0.89)	0.07
Log of family income	−0.04 (0.96)	0.03	0.00 (1.00)	0.03
Ever on public assistance	−0.31 (0.73)**	0.09	−0.07 (0.94)	0.08
Observations	14103		14113

Note: Analysis adjusts for complex survey design using STATA 9.2 svy commands. Other background controls included, not shown. Odds ratio = e^B= exponentiated B shown in parentheses under B estimate.Gay/lesbian and bisexual coefficients not significantly different from each other in any model.

^*p < 0.05;

^**p < 0.01

Table 4.

Logistic and Ordered Logistic Models of Importance of Affective Values

	Importance of lifelong commitment				Importance of love				Importance of faithfulness
	1		2		1		2		1		2
	B	SE B	B	SE B	B	SE B	B	SE B	B	SE B	B	SE B
Woman	0.59 (1.80)**	0.06			0.74 (2.10)**	0.08			0.82 (2.26)**	0.07
Gay/lesbian	−0.62 (0.54)**	0.22			−0.41 (0.67)	0.27			−0.48 (0.62)	0.27
Bisexual	−0.94 (0.39)**	0.22			−0.43 (0.65)	0.26			−0.86 (0.42)**	0.26
Straight women (a) (ref)
Straight men (b)			−0.63 (0.53)ac	0.06			−0.78 (0.46) a	0.08			−0.85 (0.43) a	0.08
Lesbians (c)			−1.38 (0.25) ab	0.25			−1.28 (0.28) a	0.33			−1.10 (0.33) a	0.26
Gay men (d)			−0.75 (0.47) a	0.28			−0.67 (0.51)	0.35			−1.02 (0.36) a	0.36
Bisexual women (e)			−1.05 (0.35) a	0.24			−0.55 (0.58)	0.31			−0.94 (0.39) a	0.27
Bisexual men (f)			−1.18 (0.31) a	0.52			−0.90 (0.41)	0.55			−1.52 (0.22) a	0.54
Observations	14109		14109		14109		14109		14111		14111

Note: Analysis adjusts for complex survey design using STATA 9.2 svy commands. All other background controls included, not shown. Odds ratio = e^B = exponentiated B shown in parentheses under B estimate. Reference category denoted by (ref). Letter superscripts ‘a–f’ indicate significant differences at p < 0.05. Gay/lesbian and bisexual coefficients not significantly different from each other in model 1 for each affective value.

^*p < 0.05;

^**p < 0.01

Figure 1.

Predicted Probabilities of Rating Affective Values “Extremely Important”

RESULTS

Table 1, Panel A shows descriptive statistics for our key independent variables with those groups serving as the reference category in later models indicated by (ref). Percentages are weighted, and N’s are unweighted. Ninety-seven percent of respondents identified as straight, and 1.5 and 1.6 percent identified as gay/lesbian and bisexual, respectively. When further defined by gender, gay men slightly outnumbered lesbians, but among bisexuals, women outnumbered men. Control variable distributions are shown in Panel B.

Table 2 shows the distributions of our relationship values overall and in cross-tabulations with gender and sexual identity. Forty-five percent of all respondents rated racial homogamy as not important but almost a quarter rated it as important to very important. In contrast, 9% of all respondents rated having enough money as not important but one-half rated it as important to very important. It appears that sexual minorities were substantially less likely than straight young adults to claim racial homogamy or having enough money as important status relationship values. Among heterosexuals and bisexuals, racial homogamy was less important to women than men, but there was less difference by gender in the importance of having enough money.

In the bottom three panels, the most important pattern is the consistently high percentages of ‘extremely important’ ratings on lifelong commitment, love, and faithfulness across all sexual identity categories. Seventy-six percent rated lifelong commitment as extremely important, 13% rated it as very important, and just 11% rated it as not important to important. Straight women were more likely than straight men to rate lifelong commitment as extremely important, but lesbians were less likely than gay men to do so. Bisexuals did not show large gender differences. Overall, straight respondents rated lifelong commitment as extremely important more often than sexual minorities. Love was rated as extremely important by 86% of all respondents. Among bisexuals and heterosexuals, women rated it as extremely important more often than men, yet lesbians rated love as somewhat less important than gay men. Straight women rated love as extremely important in an exceptionally high proportion of cases. Finally, 89% of all respondents said faithfulness is extremely important. Ratings were highest for straight women.

Tables 3 and 4 show associations expressed in terms of ordered log odds in the presence of controls for a rich array of background factors. For brevity, we show results for only a subset of background factors, though all are included in the models (full results available upon request). The ordered logit coefficients are interpreted to mean that a one unit increase in the independent variable is expected to change the dependent variable by its respective coefficient in the ordered log odds scale with the other variables in the model held constant. Because interpretation can be difficult, especially in the case of the multiple contrasts that result from significant interaction terms, we graph several key findings from Table 4 in Figure 1.

Table 3 displays results for the status trait values. On the importance of racial homogamy, coefficients show that women rated it as less important than men, and gay/lesbian and bisexual respondents rated it as less important than straight young adults. On the importance of having enough money, women placed a significantly higher value on financial security than men and gay/lesbian and bisexual respondents rated it as less important than straight respondents. We tested gender-by-sexual identity interactions for the importance of racial homogamy and having enough money, but we found no significant differences.

Table 4 shows results of models estimating associations between sexual minority status and the affective relationship values of lifelong commitment, love, and faithfulness. For each value, we include Model 2 showing associations for gender-by-sexual-identity subgroups because interaction terms indicated significant differences (not shown). We included all control variables in our analyses, but only show gender and sexual identity subgroup variables here due to space constraints. The first set of models estimates the ordered log odds of ratings of lifelong commitment. Model 1 shows that women rated lifelong commitment as more important than men, and sexual minorities rated it as less important than heterosexuals. The interaction between gender and sexual identity was significant, so we estimated Model 2 to show contrasts between gender-specific sexual identity subgroups in the ratings of lifelong commitment. Subgroups are defined by letters ‘a’ through ‘f’ in the left-most column, and these letters are used to denote statistically significant contrasts (p < 0.05). All coefficients are estimates of difference from straight women (ref); however, where differences between other groups exist, letter superscripts denote significant contrasts. Compared to straight women, all others rated lifelong commitment as less important. Lesbians rated lifelong commitment lower than all others, but they were not significantly different from other sexual minorities.

The second set of models estimates the log odds of rating love as extremely important. Model 1 shows that compared to men, women were more likely to rate love as extremely important, but there were no significant differences by sexual identity in ratings of love. Again, the interaction between gender and sexual identity was significant, so Model 2 shows contrasts between the gender-specific sexual identity subgroups. Compared to straight women, straight men and lesbians were less likely to rate love as extremely important.

The third set of models estimates the ordered log odds of ratings of faithfulness. Model 1 shows a positive association between being a woman and the importance of faithfulness and a negative association between bisexual identity and ratings of faithfulness. Next, we show the contrasts between each gender-by-sexual-identity subgroup for faithfulness. Here we see that all sexual minorities rated faithfulness as less important than straight women, but there were no differences among the other gender-by-sexual-identity groups.

To illustrate differences, we graphed the predicted probabilities of reporting ‘extremely important’ for the three affective values for gender-by-sexual-identity groups with other measures held at their means or modes. We show the affective values and not the status values because the affective values are those for which we found significant gender-by-sexual identity differences. Thus, the figure highlights these more complex patterns of difference. Letters indicate significant contrasts between groups.

First, the figure shows all subgroups, regardless of gender or sexual identity, rated the three affective values highly. Second, where there were differences across the three affective values, we found support for our hypothesis that women (but only straight women) rate lifelong commitment, love, and faithfulness more highly than men. For two of the three values—lifelong commitment and faithfulness—we also found support for our hypothesis that sexual minorities will place less importance on these values than straight respondents. Furthermore, we found support for the intersectionality hypothesis: that gender differences found among heterosexuals do not always hold among sexual minorities. In fact, the hypothesized gender differences applied only to heterosexuals. We found that lesbians’ ratings on the three values were never significantly different from those of gay or bisexual men. Although we did not find gender differences within sexual minority groups, we did find one difference between men and women across sexual identity—lesbians rated lifelong commitment as less important than straight men.

It is worth noting here that although we had just three general hypotheses, these led us to conduct a large number of statistical tests for subgroup differences. With so many tests, we run the risk of generating a significant finding when it does not really exist. We could have downwardly adjusted p-values, but we did not do so because despite our relatively large sample, we had few cases in each sexual-minority-by-gender subgroup, already compromising our ability to detect significant differences. Moreover, our findings were largely consistent with theory and past research where it exists, giving us some confidence that our significant findings are real.

DISCUSSION

Our analyses point to four main findings regarding gender, sexual identity, and young adult relationship values. The first three concern the affective values, components of the romantic love myth; the fourth concerns status values. First, we found that the romantic love myth is a pervasive cultural ideal that is held regardless of gender or sexual identity; an overwhelming proportion of respondents rated love, faithfulness, and lifelong commitment as extremely important for marriage or long-term relationships. It is worth repeating that because their distributions were highly skewed, we collapsed responses on the original measures for the affective values into a smaller set of categories, but most of those who were coded below the top category of ‘extremely important’ rated the measure as an 8 or 9 on a 10-point scale. Consistent with Swidler’s (2001) proposition, this points to the continuing cultural power of the romantic love ideal for most Americans. This finding is in contrast with Giddens’ notion of a cultural shift away from romantic to confluent love, which we would have expected to produce lower ratings for the importance of lifelong commitment and faithfulness. Although we cannot assess the trajectory of these value ratings over time, the fact that they remain high indicates that even if they have diminished recently, the descent has been minimal.

Our second main finding was that, despite these high ratings, there were significant differences by gender and sexual identity in affective relationship values: women valued love, faithfulness, and lifelong commitment more than men, and sexual minorities valued faithfulness and lifelong commitment less than heterosexuals. The gender difference in affective values applied only to heterosexuals—where we found consistent differences between straight men and women, we found no such differences between sexual minority men and women. This reflects the fact that the already high ratings on the values of love, faithfulness, and lifelong commitment were exceedingly high for straight women. In fact, although straight women were different from both straight men and all sexual minorities, straight men did not differ from most sexual minorities on affective values. This suggests the limitations of overarching gender theories (such as gender socialization theory) that fail to account for the cross-cutting influence of sexuality. Instead, a theory that recognizes the unique relationship views that arise at the intersection of these two identities affords a more accurate understanding of the multiple features of identity that shape affective relationship values.

The above finding is consistent with the intersectionality framework which implies that the values we hold for relationships are not reducible to the component parts of our identity, but instead are the product of our various identities. Among the gender theories that motivated our analyses, the theories of gender performativity and gender as a social institution, if slightly tweaked, are also compatible with these findings. The idea of gender performativity suggests that gender is emergent in social interactions and its enactment is variable and situation-dependent. Therefore, if one’s relationship is non-normative with regard to gendered coupling, then affective relationship values may be less influenced by traditional gender norms. If this is the case, individuals may perform gender differently when it comes to their relationships than they might in other domains of life. The idea of gender as an institution purports that relevant institutions, like marriage, are critical in governing gendered behavior. Therefore, if the most relevant societal institution is closed to a segment of the population, those in this segment may be less likely to adhere to gender norms reinforced by that institution. In short, some gender theories can be extended to accommodate a second (or third or more) identity category.

Third, within the broad category of sexual minorities, we found no differences on the three affective values between gays, lesbians, bisexual men, and bisexual women; all sexual minorities rated these dimensions at similar levels of importance. The fact that we found gender differences among heterosexuals but not among sexual minorities may be due to the marriage constraint shared by sexual minorities who prefer a same-sex partner. Their similar affective relationship values may reflect an adaptation to that constraint, whereas straight young adults are more likely to develop their relationship values according to gendered relationship norms that are reinforced by marriage, especially values of sexual fidelity and lifelong commitment.

The theories we reviewed did not necessarily predict gender differences among sexual minorities in ratings of components of the romantic love ideology, but some past empirical research suggested differences between lesbians and gay men. For example, some studies have found higher dissolution rates in lesbian relationships, suggesting lesbians value commitment less than gay men, but our data do not support this proposition. Also, prior research suggested that gay men practice sexual monogamy less often than lesbian and straight couples and place less importance on it. In our study, we found no significant difference in ratings of the importance of faithfulness between gay men and other sexual minorities or heterosexual men. There are several possible reasons for our inability to replicate this past finding. First, young adults may hold relationship values that they do not necessarily adhere to in practice at this stage in their life when they are exploring dating relationships more casually. Second, perhaps faithfulness has a somewhat different meaning for gay men—faithfulness may refer to the emotional primacy of the relationship rather than its sexual exclusivity (Adam, 2006). Finally, younger gay men (those represented in our study) have come of age in the wake of the HIV/AIDS crisis and therefore may hold more mainstream relationship values than cohorts of gay men who came of age before the crisis and in an era that emphasized sexual liberation and rejection of mainstream relationship beliefs.

Our fourth main finding concerns the value of status features. We found that sexual minorities were more tolerant of racial difference and less concerned with financial security in long-term relationships. This finding is consistent with Rosenfeld’s ideas about the independent life stage—growing geographic and social separation from parents has led young people to form relationships of their own choosing, rather than the more traditional pairings their parents would have chosen for them (Rosenfeld, 2007). Rosenfeld notes that the same social forces contribute to the rise of interracial and same-sex unions, and as a result same-sex relationships are more likely to be interracial. Rosenfeld’s theory suggests that sexual minorities will generally be more open to relationships that violate social norms, such as interracial relationships, because they have already violated the norm of heterosexuality and achieved considerable independence from their parents’ expectations and influence. An alternative explanation is that sexual minorities have a smaller pool of eligible partners and thus must be more accepting of racial difference or financial instability if they desire a relationship (Ellingson et al., 2004).

Taken together, our main findings have several important theoretical implications. Theorizations of gender that can accommodate the cross-cutting influence of other identity categories such as sexual orientation are best suited to account for the role of gender in formation of relationship views. Middle-range theories positing sexual minorities as relationship innovators have some for explaining some differences in relationship views (i.e., on status characteristics such as racial homogamy and financial security) but run the risk of exaggerating differences between sexual minorities and heterosexuals on other dimensions (i.e., affective relationship values). The intersectionality framework provides an especially useful lens for making sense of the complex relationship between identity characteristics and relationship views. Our study provides some quantitative confirmation for previous qualitative work emphasizing the broad power of cultural myths such as the romantic love ideal, which appears to transcend differences of gender and sexual identity among young adults.

Our findings also point to several important areas for future research on gender, sexual identity and relationship views. First, research should attempt to gauge the specific mechanisms linking gender, sexual identity, and relationship values, and adjudicate among competing explanations (e.g., confluent love v. incomplete institutionalization). Second, although we tested the associations between relationship values and the gender-sexual identity nexus, we would have liked to test the contingent influence of additional identity categories like race and class on relationship values. Unfortunately, we have too few cases in most race- or class- and gender-specific sexual identity groups. If possible, future research should investigate the intersection between a broader set of identity categories and the valuation of partnership characteristics. Third, researchers would benefit from data that tracks the relationship views of sexual minorities of different cohorts over time and across place. This would allow examination of age, cohort, and geographically-specific legal influences on relationship views to test whether views change over the life course or with the shifting social, legal and cultural context. Finally, future research should explore the interaction between gender and sexual identity among young adults to confirm or refute our findings that lesbians place less importance on lifelong commitment and straight women place more importance on lifelong commitment, love, and faithfulness than most other young adults.

This study shows that among recent cohorts, women still value affective relationship dimensions more than men, and sexual minorities hold different relationship views than heterosexuals on some dimensions: they are more accepting of racial difference and less concerned with financial security. In other ways sexual minority young adults are quite similar to heterosexual young adults: they highly value affective relationship traits like lifelong commitment, love, and faithfulness. Among young adults, most sexual minorities are no different than straight men in their rating of the importance of lifelong commitment, love, and faithfulness. All sexual minority groups are different than straight women on ratings of lifelong commitment and faithfulness, but so are straight men. Rather than indicating a division along sexual identity lines, these findings highlight the exceptionally high value that straight women place on these traits. Although we observe small intergroup differences, we submit that an equally important story is the overall high ratings on the values of love, faithfulness, and lifelong commitment across sexual identity groups, a pattern that suggests that the romantic love myth remains highly influential as this new generation of young adults comes of age.