A large majority of young adults (18–24-year-olds) in the United States are sexually active.^1,2 Furthermore, many adolescents and young adults have sexual partners with whom they are not in a committed romantic relationship^3–8—often referred to in the popular media as "friends with benefits" or "hookups." For example, in one study, 38% of sexually active adolescents had had sexual intercourse in the previous year with someone they were not dating;⁶ in another, 22% of unmarried college students reported that their last sexual encounter was with a "casual" partner.⁵ In a study of undergraduate students, 78% had experienced some sexual activity with a brief acquaintance or a stranger during their college years; among those, 48% of males and 33% of females had had intercourse as part of those experiences.³ Interestingly, a study of community college students found that various sexual milestones occurred earlier with a casual partner than with a committed partner.⁹

The phenomenon of friends with benefits among young people has been highlighted in the media^10,11 and has engendered speculation that such sexual partnerships may be emotionally damaging.¹² Traditional social mores dictate that sexual expression be "saved" for a romantic partner, and for more than a decade abstinence-only education policy required teaching that "a mutually faithful monogamous relationship in the context of marriage is the expected standard of human sexual activity," and that "sexual activity outside of the context of marriage is likely to have harmful psychological and physical effects."¹³ Departure from these norms and teachings may generate emotional conflict for some young people, which could vary by group norms linked to race or ethnicity, socioeconomic status or other characteristics. Likewise, physical intimacy without concomitant emotional intimacy may leave one feeling used, feeling unable to attract a romantic (not merely sexual) partner or questioning one's self-worth; any of these outcomes may jeopardize psychological well-being.

Nevertheless, few peer-reviewed studies have examined the mental health associations with (or consequences of) nonromantic sexual activity.^3,4,7 Grello and colleagues found that among middle and older adolescents, the transition from being sexually inexperienced to engaging in casual sex was associated with greater psychological distress than the transition from being sexually inexperienced to engaging in romantic sex.⁴ However, they noted that depressive symptoms were already present among adolescents who transitioned to casual sex, suggesting that such sexual activity may be an indicator of psychological distress rather than its cause. In a more recent study, the same team found that casual sex (but not romantic sex) was associated with symptoms of depression among female college students, but that males who engaged in casual sex had the fewest depressive symptoms of any group studied.⁷ Depressive symptoms have also been found to predict sexual debut and high-risk sexual behaviors among adolescents, irrespective of partner type,^14–16 and to be exacerbated by "romantic involvement," particularly among females.¹⁷

Several reasons may account for gender differences in casual sex and in the observed associations between casual sex and psychological well-being. First, the finding that males report significantly more hookups, friends with benefits and casual sex^3,7 may stem from a disposition among young women to characterize a relationship as more substantial than it actually is in the hope that it will blossom into a romantic partnership.⁷ Second, although a growing body of literature suggests that adolescent females may be the sexual initiators in many cases,^18,19 the noted gender differences may reflect a broader compliance with social norms that continue to view casual sex as acceptable for men but inappropriate for women.²⁰ And third, psychological distress may precede involvement in casual sexual activity, and depressed young women may seek external validation from or intimacy with a sexual partner because of negative feelings of self-worth or isolation.⁷

The purpose of the present study is to examine types of sexual partnerships among young adults, and to compare the prevalence of casual partnerships across demographic and personal characteristics. In addition, we assess four elements of psychological well-being across groups reporting different types of sexual partners. On the basis of previous research findings, we expect to find stronger inverse associations between casual partnerships and psychological well-being among females than among males. Because earlier research on casual partnerships has focused primarily on white college students and used depressive symptoms as the only major dependent variable, this study extends the existing literature by using a large, diverse sample of sexually active young adults, which allows us to extrapolate to a broad cross section of this population.

METHODS

Data for this study are from Project EAT-II, a follow-up study of the social, environmental, personal and behavioral determinants of weight status and eating behavior among adolescents. In Project EAT-I, a sample of 4,746 middle and high school students from 31 Minnesota public schools completed in-class surveys during the 1998–1999 academic year; Project EAT-II resurveyed these original participants by mail five years later (April 2003 to June 2004). A total of 2,516 adolescents (68% of those for whom we had accurate contact information at follow-up) participated in both waves. The current analysis was restricted to participants who had been in high school during the first survey, as only these individuals were asked about sexual behaviors on the second survey. This subsample included 1,710 young adults (764 males and 946 females), whose mean age at follow-up was 20.4 years (standard deviation, 0.8). Additional details about Project EAT are available elsewhere.^21,22

The original survey was revised to be appropriate for young adult participants in Project EAT-II. Surveys were mailed to addresses provided by participants at baseline, and Internet tracking services were used to identify correct addresses when mail was returned as undeliverable. Nonresponders were sent two reminder postcards and three additional survey packets. Completion of the survey implied consent to participate, and all study protocols were approved by the University of Minnesota's institutional review board.

Measures

•Partner type. Respondents were asked to describe their most recent sexual partner, using one of the following categories: stranger; casual acquaintance; close but not exclusive partner; exclusive dating partner; fiancé(e), spouse or spousal equivalent; or other.² Those who indicated that they were not sexually active (348) or who responded "other" (22) were excluded from the analysis. In addition, 29 respondents who said their most recent partner was a stranger were excluded because of the possibility that these experiences were nonconsensual. These exclusions reduced the analytic sample to 1,311 (574 males and 737 females; mean age, 20.5). Comparisons were made across the remaining four partner categories, as well as by the dichotomous classification of "casual partner" (casual acquaintance or close but nonexclusive partner) versus "committed partner" (exclusive partner, fiancé or fiancée, spouse or spousal equivalent).

•Psychological well-being. Body satisfaction, self-esteem and depressive symptoms were assessed using psychometrically valid and reliable scales, on which the higher the score, the greater the characteristic being measured. Body satisfaction was assessed using a five-point Likert scale of satisfaction with 10 body parts or attributes²³ (range, 10–50; Cronbach's alpha, 0.93 for males and 0.92 for females). Self-esteem was measured using a four-point Likert scale for six items from the Rosenberg Self-Esteem Scale,²⁴ including such items as "On the whole, I am satisfied with myself" (range, 6–24; Cronbach's alpha, 0.82 for both males and females). Six depressive symptoms were assessed, such as feeling hopeless about the future and feeling unhappy, sad or depressed;²⁵ response options were "not at all,""somewhat" and "very much" (range, 6–18; Cronbach's alpha, 0.80 for males and 0.81 for females). Finally, suicidal ideation was measured with the question "Have you ever thought about killing yourself?" Those who responded "yes, during the past year" were compared with those who responded "yes, more than a year ago" or "no," to approximate the same time frame as used for the other indicators of psychological well-being (i.e., current or past year).

•Demographic and personal. Respondents’ race or ethnicity was assessed as white, black, Hispanic or Latino, Asian-American, Native American or American Indian, or other; respondents were asked to check all that applied, and those indicating more than one category were grouped with "other." Five levels of socioeconomic status were based primarily on the highest education level completed by either parent (ranging from "did not finish high school" to "master's degree or Ph.D."). In cases where this information was missing, eligibility for public assistance, eligibility for free or reduced-cost school meals, and parental employment status were used to infer socioeconomic status.^26-27 Student status over the past 12 months was categorized into three groups: full-time at a four-year college, part-time at a four-year college or attending a community or technical college, and not in school. Sexual orientation was assessed by the question "Which of the following best describes your sexual orientation?"; response options were "attracted to opposite gender,""attracted to same gender,""attracted to both genders" and "questioning." Those attracted to the opposite gender were compared with all other groups combined.

Analysis

Chi-square tests were used to compare the prevalence of casual partnerships (dichotomous variable) by participants’ demographic and personal characteristics. General linear modeling was used to compare mean levels of each psychological well-being measure between the casual and committed partner variables, as well as across the four partner categories. Two models were run for each psychological measure: The first was unadjusted, and the second was adjusted for demographic and personal characteristics and the baseline level (assessed in Project EAT-I) of the relevant well-being variable. Least-square means generated by general linear modeling were interpreted as predicted probabilities of a positive response on the dichotomous suicidal ideation variable. To fully examine differences between each partner category, post hoc tests were run even when the overall F statistics were nonsignificant; however, these findings should be considered exploratory. In addition, unadjusted analyses using the casual versus committed partner variable were performed, first restricting the sample to college students and then to white participants, to facilitate comparisons with previous research on this topic. All analyses were stratified by gender and were conducted with SAS version 8.2.

RESULTS

Two-thirds of respondents were white, and the sample was well distributed across socioeconomic levels (Table 1). Thirty-seven percent of participants were full-time students at a four-year college, and 27% were part-time students at such colleges or were enrolled at a community or technical college; 37% were not in school. Most respondents (94%) said they were attracted only to the opposite sex.

Fifty-five percent of respondents reported that their most recent sexual partner was an exclusive dating partner, while 25% said their last partner was a fiancé(e), spouse or spousal equivalent. Twelve percent reported that their last partner was a close but not exclusive partner, and 8% said the partner was a casual acquaintance. Mean scores for body satisfaction, self-esteem and presence of depressive symptoms were generally midrange, and 16% of respondents said they had thought about killing themselves in the past year.

Twenty-nine percent of males and 14% of females reported that their last sexual partner was casual (chi-square=44.8, p&;.001; not shown). The proportions of male and female respondents reporting a recent casual sex partner differed by race or ethnicity: Percentages ranged from 16% of Asian-American men to 58% of men who indicated they were of "other" racial or ethnic background; among women, percentages ranged from 5% of Asian-Americans to 36% of Native Americans (Table 2). No significant differences were found among males or females by socioeconomic status, student status or sexual orientation. Reporting of a recent casual partner appeared to be more frequent among men who were attracted to men or to both genders, or who were questioning, than among men who were attracted only to women (45% vs. 29%), but this difference was not significant.

Psychological well-being scores were notably similar across sex partner categories, spanning less than one point in many cases (Table 3). For instance, the mean self-esteem scores ranged from 19.1 to 19.7 for men and from 17.2 to 17.8 for women across the four sexual partner categories in the adjusted models; narrow ranges were also found for body satisfaction and presence of depressive symptoms. The only significant model was the unadjusted analysis that compared males’ reports of depressive symptoms: Men with committed recent partners reported slightly more symptoms than did those with casual partners (11.9 vs. 11.2). However, this association was no longer significant after we controlled for covariates.

Although the four-level partner type variable was not associated with psychological well-being in any model, post hoc tests revealed significant differences between certain partner categories. Compared with men whose most recent partner was a casual acquaintance or an exclusive dating partner, men whose recent partner was a close but not exclusive partner were more likely to have thought about suicide in the previous year (15% vs. 4–7%), after adjustment for covariates. Women whose most recent sex partner was a fiancé, spouse or spousal equivalent reported fewer depressive symptoms than those whose last partner was an exclusive dating partner (11.6 vs. 12.2).

Finally, our analyses that were restricted to college students and to white participants produced no statistically significant findings.

DISCUSSION

In spite of speculation that casual partnerships could be psychologically damaging for sexually active young people, our findings show almost no differences in psychological well-being between those with a casual partner and those with a more committed partner; these results were similar in the analyses using the dichotomous variable and the four-category partner variable. Interestingly, in the few instances in which post hoc tests detected significant differences, associations did not reflect a smooth "dose-response" relationship, whereby psychological well-being was incrementally worse in successively casual types of partnership. Rather, intermediate categories (exclusive dating partner and close but not exclusive partner) were linked to poorer well-being among men in some cases. Only in the analysis of women's depressive symptoms was the most committed partner category associated with better mental health. Sexual partnerships in intermediate categories may entail stressors aside from their level of "casualness" that are relevant to psychological well-being. Findings from tests of difference between specific categories should be interpreted with caution, however, because of the large number of comparisons made across groups, and the fact that the core construct was not associated with psychological well-being in these adjusted models.

Previous studies have found a higher prevalence of casual partnerships among young adults than we did.^3,7 This difference is likely due to the assessment of casual sexual partnerships "ever" or during the college years, whereas our analysis focused on participants’ most recent sexual experience. Additionally earlier studies have found casual sexual behavior to be more common among males than among females (as we did),^3,7 and more common among those with same-sex experience than among others (as our results suggested).²⁸

However, overall we found no associations between type of recent sexual partner and indices of psychological well-being in our fully adjusted analyses, whereas other research has found that young adult females reporting casual sex had greater depression and males reporting casual sex had less depression than those in other groups.⁷ Several differences in research design may account for this disparity. First, previous studies have relied largely on samples of white college students,^3,7 for whom behaviors and associations may differ from those of a more socially diverse sample. Yet our supplemental analyses restricted to college students and to white participants also revealed no associations between partner type and psychological well-being. This suggests that the null findings in the full analysis are not merely a reflection of a particular pattern of results in a more diverse sample. Second, earlier work focused exclusively on issues of sexual partnerships and psychological well-being,⁷ and may have drawn a sample for whom these issues were particularly salient; in contrast, our study focused largely on unrelated issues (nutrition, eating behaviors), and so may have captured a broader sample without well-defined thoughts about sexual behavior.

Another potential explanation of the present study's predominantly null findings is that we specifically assessed participants’ most recent sexual partner. Yet sexually active young people typically report both romantic and casual partners in their sexual history.⁴ Therefore, the distinction between "casual partner" and "committed partner" groups may not adequately differentiate participants; indeed, which type of partner happened to be the most recent may be a matter of chance. This misclassification would bias associations toward the null. However, the second analysis, using four partner categories, also found no associations after adjusting for relevant covariates. Partner misclassification is less likely for those who are engaged or married, which tempers our concern about bias toward the null. The finding that those with a fiancé(e) or spouse as their most recent partner differed significantly from other groups in post hoc tests for only one adjusted model lends credibility to the interpretation of a genuine null association.

Our findings do not minimize the legitimate threats to physical well-being associated with casual sexual relationships, and the need for preventive messages in sex education programs and other interventions with young adults. Most research on the topic has shown that condom use is more common with casual than with primary partners. This association is particularly pronounced in studies of young people's sexual behaviors in Africa and Asia, although it has been found in U.S.- and European-based research as well.^29–36 Nonetheless, inconsistent condom use (except when partners are known to be disease-free or when other effective forms of protection are used) constitutes an ongoing public health and clinical concern: Nationally, only 53% of males and 31% of females aged 20–24 report using a condom at their last sexual encounter.¹ In particular, disease status is more likely to be unknown for casual partners than for long-term partners, and hence the risk of infection may be greater in casual partnerships. In addition, positive attitudes about casual sex have been associated with a history of sexual aggression among men and attitudes conducive to intimate partner violence.³⁷ Therefore, women who engage in casual sexual relationships may also be at risk of physical harm from their partners.

Limitations and Strengths

This study has several limitations that should be considered when interpreting the findings. First, because baseline sexual activity measures were not available, we were unable to follow respondents with different sexual partnerships over time to detect changes in their psychological well-being—an important step in identifying a causal relationship. Second, a single item was used to categorize sexual partnerships. Other potentially important information—such as the nature of participants’ first sexual partnerships, frequency of casual versus committed sexual activity, and beliefs about casual sexual activity—was not available. Third, our measure of socioeconomic status relied primarily on information about participants’ parents, rather than about participants themselves. Income data were not collected for Project EAT, as they tend to be unreliable when reported by young adolescents. Rather, we included a measure of participants’ student status to enhance our assessment of their socioeconomic position.

Finally, although the response rate among those who were contacted for Project EAT-II was good, participants in the follow-up sample were more likely than those in the original sample to be white (68% vs. 55%) and in the two highest socioeconomic categories (43% vs. 40%). These differences are likely due to the greater mobility of participants in the lower socioeconomic and nonwhite groups, who were heavily represented in the baseline sample. Because sexual activity items were added to the survey at follow-up, we were unable to determine if attrition was related to sexual behavior. However, since the primary subject matter of the survey was nutrition and weight concerns, we expect that sexual behavior would be unrelated to a decision to participate at follow-up, and would therefore not introduce bias into our analysis.

One of the main strengths of this study was the use of a large, population-based sample that was diverse in terms of race and ethnicity, socioeconomic status, student status and sexual orientation. The study was adequately powered, reducing the possibility that the null findings were due to type II error rather than a genuine lack of association. In addition, the study relied on established, psychometrically sound measures of psychological well-being, which have been used extensively in research with young people, providing a broader picture of emotional health than has been examined previously. Finally, the use of both a dichotomous and a four-level categorical sexual partner variable was valuable. The dichotomous variable assured adequate power to detect even relatively small differences between groups, while the four-category measure gave greater precision in describing types of sexual partnerships (although each category necessarily included a smaller number of participants). The fact that the results concurred for the two forms of the partner variable suggests the robustness of the predominantly null relationships between partner type and psychological indices.

Conclusions

Our findings indicate that the large majority of sexually active young adults are involved in committed or exclusive relationships, and that those in casual relationships do not appear to be at increased risk for harmful psychological outcomes. These findings, while reassuring, should not suggest that interventions aimed at promoting safer-sex behaviors are unnecessary. Interventions and health education focusing on health risks from casual sexual encounters continue to be appropriate for this age-group and for adolescents, particularly given the inconsistency of condom use in general, and dual-method contraception in particular.³⁸ Indeed, in 2008 the Centers for Disease Control and Prevention reported a 26% STD prevalence among women aged 14–19.³⁹ Furthermore, regardless of whether sexual relations occur in a casual or a more exclusive relationship, young adults may need guidance in ensuring that they are adequately prepared, at a psychological level, to have open communication with their partner.

Future research should explore the progression of sexual activity choices and variations in psychological well-being over time using multiple assessments, and should include measures that can assess various characteristics of sexual relationships, such as duration of partnership, frequency of sexual activity and presence of violence. In addition, understanding qualitative aspects of sexual relationships may help us to identify characteristics that are associated with health risks, and consequently to develop more effective, targeted interventions for youth.