Changes in Diverse Sexual and Contraceptive Behaviors Across College.

Researchers have made repeated calls for a better understanding of normative sexuality development during adolescence and young adulthood. We examined how the occurrence of seven penetrative, nonpenetrative, and contraceptive behaviors changed longitudinally across seven waves, and how individual (gender) and contextual (romantic relationship status) factors related to these changes in a sample of college students (N = 730, M age = 18.4 at Semester 1; 51% female; 26% Hispanic/Latino American, 22% Black/African American, 30% Asian American/Pacific Islander, 45% White/European American). Across college, reported kissing, touching, performing and receiving oral sex, and penetrative sex rates increased, and contraception use (any type) and condom use (in particular) rates decreased, demonstrating changes with age independent of young adults' romantic relationship experiences. Rates of all sexual behaviors were higher, and of contraception use lower, when students were in serious romantic relationships. Contraception use decreased more for men than for women, particularly in semesters men were not in serious relationships. Condom use decreased for men, and for women in semesters they were in serious relationships. Findings demonstrate normative trends in sexuality development, as well as suggesting the value of enhanced sexual health promotion programming, with a particular focus on contraceptive behaviors, across college.

Understanding sexual and contraceptive behaviors during the developmental period of young adulthood is conceptually complex because these behaviors are developmentally normative, but still incur potentially life-changing health risks. The average age at first vaginal sex is 17 (Centers for Disease Control and Prevention, [7]); the average age at first marriage is 27–29 (U.S. Census Bureau, [52]). As a result, for traditionally aged students, the college years (approximately age 18–23) correspond to a developmental period when the majority of individuals are sexually active and unmarried. Young adults in general, and college students in particular, often experience substantial exploration in a range of domains, including sexuality (Arnett, [3]; Lefkowitz & Gillen, [26]). Thus, researchers have called for a better understanding of normative sexuality development during adolescence and young adulthood (Russell, [44]; Tolman & McClelland, [51]).

Sexuality scholars have outlined the value of a normative developmental framework on sexuality. First, sexuality is a normative task. That is, it is typical, appropriate, and healthy to express interest in and explore sexuality during adolescence and young adulthood (Tolman & McClelland, [51]), when establishing romantic and sexual relationships is an important task (Arnett, [3]). Recognition of sexuality as normative shifts the focus of research solely from identifying "risk factors" for engaging in sexual behavior to understanding the experiences of sexual behaviors, including behaviors like kissing and touching that do not confer the same physical health risks as vaginal and anal sex. Second, a normative developmental framework recognizes that individual and contextual factors such as gender and romantic relationships influence and shape sexual experiences (Russell, [44]). Third, sexuality is a developmental phenomenon, in that it changes across adolescence and young adulthood with both age and context. Finally, although sexual behaviors are associated with negative outcomes such as sexually transmitted infections (STIs) and unwanted pregnancy (Lefkowitz & Gillen, [26]), research also demonstrates beneficial associations with physical, mental, and relationship health (Ebrahim et al., [13]; Nicolosi, Moreira, Villa, & Glasser, [32]; Sprecher, [50]; Vasilenko, Lefkowitz, & Welsh, [54]). Therefore, basic knowledge about sexual and contraceptive behaviors and how they change with time is necessary for understanding a central domain of human development that is normative and potentially both risky and health-enhancing.

Although most work on sexual and contraceptive behaviors and health focuses on penetrative sex and condom use, other sexual and contraceptive behaviors also have implications for well-being. Young adults often perceive oral sex as a positive part of their sexuality, playing important roles in physical pleasure and relationship intimacy (Lefkowitz, Vasilenko, & Leavitt, [27]). Similarly, college students are more positive about first kisses than about any other first sexual behaviors (Vasilenko, Maas, & Lefkowitz, [55]) and kissing contributes to adolescent couples' healthy relationship functioning (Welsh, Haugen, Widman, Darling, & Grello, [56]). Thus, establishment of healthy nonpenetrative, penetrative, and contraceptive behaviors in young adulthood can contribute to short- and long-term physical, mental, and relationship health by predicting concurrent well-being and subsequent healthy sexuality in adulthood.

In this paper, we study normative sexuality development in young adulthood by documenting how the recent occurrence of seven penetrative, nonpenetrative, and contraceptive behaviors changes across the college years using seven waves of longitudinal data in a racially and ethnically diverse sample of traditionally aged college students, and examine how individual (gender) and contextual (romantic relationship status) factors relate to these changes. We utilize a sample from a large state university in a relatively small college town (90,000 inhabitants, of whom half are students), and where the majority of students live on campus. This context does not apply to all college students' experience, but does share similar elements to the experience of many students who attend public colleges and universities outside major cities.

Most individuals engage in sexual behaviors before young adulthood. For example, the vast majority of individuals have kissed and touched a partner by traditional college age, that is, about 18 years (O'Sullivan, Cheng, Harris, & Brooks-Gunn, [33]; Regan, Durvasula, Howell, Ureño, & Rea, [42]; Smiler, Frankel, & Savin-Williams, [49]; Welsh et al., [56]). About two thirds have had oral sex by this age, with slightly higher proportions reporting receiving than performing (Chandra, Mosher, & Copen, [8]; Garcia, Cavalie, Goins, & King, [17]; Halpern & Haydon, [19]; Herbenick et al., [22]; Lindberg, Jones, & Santelli, [28]). Lifetime rates of vaginal sex are similar to rates of oral sex in college student and young adult samples (American College Health Association, [2]; Halpern & Haydon, [19]; Herbenick et al., [22]; Lindberg et al., [28]; Mosher, Chandra, & Jones, [31]). Late adolescence and young adulthood are also periods of substantial sexual risk taking. Although vaginal sexual behavior without any form of contraception is relatively uncommon, most sexually active students have experienced sex without a condom in their lifetime (American College Health Association, [2]).

However, despite high rates of young adults' lifetime sexual behaviors, less is known about current or more recent behaviors. Recent sexual behaviors are more informative than lifetime behaviors, because they reflect typical and ongoing experiences, including possible benefits and risks. In the current paper, we focus on sexual and contraceptive behaviors in the past three months. Notably, recent or daily rates are substantially lower than lifetime rates (Herbenick et al., [22]; Patrick, Maggs, & Lefkowitz, [37]). For example, kissing (46%) and touching (36%) a partner in the past two weeks are still relatively common during college (Patrick et al., [37]). About half of college students have engaged in oral sex, 42–68% in vaginal sex, and 5% in anal sex in the past month (American College Health Association, [2]; Oswalt & Wyatt, [34]). In terms of contraceptive use, it is relatively common for college students to have engaged in sexual behaviors without a condom in the past month (Oswalt & Wyatt, [34]).

Nearly all the research in this area has used cross-sectional data, which makes it difficult to understand whether and how sexual and contraceptive behaviors change across college and young adulthood. Cross-sectional studies suggest that both lifetime and recent rates of sexual behaviors increase from early adolescence to young adulthood (Halpern & Haydon, [19]; Herbenick et al., [22]; Lindberg et al., [28]; Mosher et al., [31]; Smiler et al., [49]; Williams, Connolly, & Cribbie, [59]). With respect to differences in any contraception use by age, cross-sectional studies have mixed results, with some suggesting more reliable contraception use among younger adolescents (Wu, Meldrum, Dozier, Stanwood, & Fiscella, [60]), others among young adults (Santelli et al., [46]; Whitaker, Sisco, Tomlinson, Dude, & Martins, [58]), and still others observing no age differences in women's use of contraception (Manlove et al., [29]). More longitudinal data exist on condom use in particular, suggesting that consistent condom use decreases from the end of high school into young adulthood (Bailey, Haggerty, White, & Catalano, [4]; Fergus, Zimmerman, & Caldwell, [14]; Lam & Lefkowitz, [25]).

To truly understand changes in sexual and contraceptive behavior across this period of development, we need longitudinal research with more frequent assessment than is typical of many panel studies (e.g., National Longitudinal Study of Adolescent Health; Harris et al., [21]). Rapid changes occur across college. At the start, residential students live independently for the first time (Schulenberg & Maggs, [47]), with decreased parental monitoring and myriad new opportunities for exploration in romantic and sexual relationships. By the end, students move toward more adult and longer-term romantic relationships, establishing careers, and making independent decisions about life choices. To capture important changes in sexual and contraceptive behavior, longitudinal data on individual change with sufficiently frequent measurement of diverse sexual and contraceptive behaviors is necessary. The current study follows students across four years of college, considers behaviors rarely examined together (e.g., nonpenetrative behaviors, performing and receiving oral sex, any contraceptive behaviors and condom use), and uses semi-annual assessments. This design helps to rule out between-person differences and provides greater precision in assessing these dynamic phenomena.

American cultural expectations set up a sexual double standard, with men expected to desire frequent sex and women expected to serve as sexual gatekeepers (Bordini & Sperb, [5]). Empirical research primarily based on cross-sectional data has not consistently observed that behavior conforms to such a double standard. Lifetime rates of kissing and touching are similarly high for male and female adolescents and college students (Garcia et al., [17]; O'Sullivan et al., [33]; Regan et al., [42]), but reported daily rates are higher for female than male students (Patrick et al., [37]). Research on gender differences in oral sex is mixed (Garcia et al., [17]; Herbenick et al., [22]; Lindberg et al., [28]; Prinstein, Meade, & Cohen, [41]). However, the majority of oral sex research does not distinguish performing from receiving oral sex, which may explain the failure to capture differences. National data suggest minimal gender differences in lifetime rates of performing or receiving oral sex, but that recent behavior differs more, with young men more likely to receive than perform oral sex, and women more likely to perform than receive oral sex (Herbenick et al., [22]; Vannier & O'Sullivan, [53]). For penetrative sexual behaviors, men tend to report higher lifetime and recent rates of vaginal and anal sex (Petersen & Hyde, [39]), with some conflicting evidence (Herbenick et al., [22]). Although limited, some research suggests that heterosexual male and female adolescents report similar rates of lifetime and recent anal sex (Lindberg et al., [28]; Oswalt & Wyatt, [34]).

In terms of contraceptive behaviors, some research suggests that male young adults and adults report more consistent condom use than women (Bailey et al., [4]; Petersen & Hyde, [39]; Pingel et al., [40]). Although less is known about gender differences in other methods of contraception, some research demonstrates similar reported rates of couple-level oral contraception use for college men and women at first and most recent sex (Siegel, Klein, & Roghmann, [48]), whereas other research demonstrates that adolescent girls report higher rates of oral contraception use than boys (Cavazos-Rehg et al., [6]).

Scholars have suggested that gender differences in rates of sexual behaviors are due to reporting issues, including the desire to conform to gender norms influencing biased recall of behavior (Alexander & Fisher, [1]). However, gender differences in performing and receiving oral sex, which involve different roles for each partner, are more feasible. Cultural scripts that provide differing sexual standards for men and women (Bordini & Sperb, [5]) may encourage men to take on the receiver and women the performer role during oral sex, translating to gender differences. The current study addressed gaps in prior literature by (a) examining gender differences with frequent assessment of recent behaviors which are less subject to recall bias, (b) examining performing and receiving oral sex separately, and (c) including use of both condoms and contraception beyond condoms, to include more female-focused contraceptive behavior.

Across adolescence, romantic relationships become an increasingly important context for development in general (Collins, Welsh, & Furman, [10]), and the development of sexual and contraceptive behaviors in particular. Scholars consider sexual behavior a key feature of romantic relationships (Collins et al., [10]). Some researchers have found that sexual behaviors outside of romantic relationships are linked to more negative mental health outcomes than sexual behaviors within relationships (Grello, Welsh, & Harper, [18]). Cross-sectional work demonstrates that college students' nonrelationship experiences often do not involve penetrative sex (Wesche, Vasilenko, & Lefkowitz, [57]). Adolescents and young adults who are not in romantic relationships are more likely to use condoms consistently, and less likely to use no form of contraception, than individuals in romantic relationships (Bailey et al., [4]; Ott, Adler, Millstein, Tschann, & Ellen, [35]; Pingel et al., [40]). In fact, condom use declines as early as three weeks after a romantic relationship starts (Fortenberry, Tu, Harezlak, Katz, & Orr, [16]). However, condom use also declines from ages 15 to 18 regardless of relationship status (Pingel et al., [40]). It remains unclear whether these decreases over time in condom use are due to age, or to changes in relationship likelihood or intensity. Given the importance of romantic relationships in this developmental period (Collins et al., [10]), the current study uses a within-person design to examine how moving in and out of romantic relationships is associated with sexual and contraceptive behavior above and beyond changes with age.

In summary, the purpose of this study was to examine changes in rates of recent (three months) sexual and contraceptive behaviors across college, and how such changes differ by gender and relationship status. Specifically, we had four primary aims: (a) to describe patterns of change in rates of kissing, sexual touching, oral sex, penetrative sex, any contraception use, and condom use across four years of college, and to examine how these longitudinal patterns are moderated by (b) gender, (c) romantic relationship status, and (d) the interaction of gender and romantic relationship status.

First-year students at a large state university received e-mail invitations to the University Life Study, a seven-semester study of college student health behaviors. Eligible individuals were less than 21 years old, U.S. citizens or permanent residents, and living within 25 miles of campus. Based on registrar records, which only recorded one ethnic or racial group per student, the student population was predominantly White/European American (84%), with 6% Hispanic/Latino, 5% Black/African American, and 6% Asian American/Pacific Islander (sums to more than 100% due to rounding). We used stratified random sampling with replacement to achieve a diverse sample with respect to gender and race or ethnicity. In total, 744 students (65%) provided informed consent and participated in Semester 1 (S1). As reported elsewhere, students who participated did not differ from those who did not on race or ethnicity or age, but were more likely to be female (Patrick, Morgan, Maggs, & Lefkowitz, [38]). Participant retention was strong (82–88%) through Semester 7 (S7, fourth year of college). Because multilevel models allow for incomplete data, we retained all measurement occasions where students had data on variables of interest, dropping 14 participants due to missing data. Participants in the analytic sample (N = 730; 51% female) averaged 18.4 (SD = 0.4) years of age at S1. Twenty-six percent of the sample identified as Hispanic/Latino American ethnicity, 22% identified as Black/African American, 30% as Asian American/Pacific Islander, and 45% as White/European American, with participants allowed to select more than one race or ethnicity. At S1, 98% of the sample lived on campus, 1% lived with parents, and < 1% lived off campus but not with parents. At S1, 98% of the sample identified as heterosexual, 1% as bisexual, < 1% as gay or lesbian, and < 1% as other. By S7, 95% identified as heterosexual. We included participants of all sexual orientations or identities because the majority of sexual behaviors assessed are not specific to a particular sexual orientation or gender pairing. This study was approved by Pennsylvania State University's Institutional Review Board.

Each semester, participants received an e-mail with a secure survey link. Participants completed informed consent electronically. Each semester, they completed one longer Web-based survey, and up to 14 days of shorter daily Web-based surveys (daily data not used in current paper). They received $20 (S1) to $40 (S7) for completing the semester surveys.

At S1, participants answered questions about participation in six different sexual behaviors (kissing, sexual touching, performing oral sex, receiving oral sex, vaginal sex, anal sex) in their lifetime. At each subsequent semester, participants only received each lifetime question if they had previously answered "no." Once participants reported lifetime occurrence of a given behavior, subsequent semester surveys asked questions about sexual behaviors in the past 12 weeks (three months). The kissing question was: "In the past 12 weeks, have you and a partner kissed on the lips?" Wording for other behaviors was: "Have you and a partner touched each other under your clothing or with no clothes on?"; "Have you performed oral sex on a partner?"; "Has a partner performed oral sex on you?"; "Have you had vaginal and/or anal sex with a partner?" The penetrative sex question combined vaginal and anal sex into one question based on prior research (American College Health Association, [2]; Oswalt & Wyatt, [34]) and our pilot data suggesting that anal sex was infrequent in this population. We used these five recent behavior questions assessed in each of seven semesters to indicate recent behavior.

Participants who reported recently engaging in penetrative sex were also asked, "In the past 12 weeks, how frequently have you used a condom when you had vaginal and/or anal sex?" and, "In the past 12 weeks, how frequently did you use any method to prevent pregnancy or disease, including condoms, when you had vaginal and/or anal sex?" For these two questions, response options were: never = 1, some of the time = 2, most of the time = 3, every time except once = 4; every time = 5. The contraception variable included any type of contraceptive use, including condoms. Following work by Bailey et al. ([4]), we dichotomized these behaviors to create two partially overlapping variables, condom use and any contraception use, with 1 indicating always using a condom (any contraception) and 0 indicating not using a condom (any contraception) at least once. We dichotomized as consistent use versus any nonuse because a single occasion of unprotected sex increases the risk of STIs and pregnancy.

At S1, participants reported whether they were male (= 1) or female (= 0).

Each semester, participants reported their romantic relationship status. We used participant responses to create two categories: in a serious romantic relationship (= 1, in a serious and committed relationship [which varied from 25–32% across the seven semesters], living with [< 1–3%], engaged [< 1–2%], or married [< 1–1%]) versus not in a serious relationship (= 0, casually dating [10–13%] or not dating anyone [52–62%]). We categorized casually dating as not being in a serious relationship because (a) sex in casual relationships is often considered to be nonrelationship sex (e.g., Fielder, Carey, & Carey, [15]; Owen & Fincham, [36]), and (b) in our sample, relationships reported during semesters when students were casually dating tended to be shorter (F = 53.19, p < .001), and were less likely to be exclusive (χ2 = 14.22, p < .001), than relationships in semesters when students were in a serious relationship.

Because sexual and contraceptive behaviors vary by age, race or ethnicity, and socioeconomic status (SES), we controlled for these sociodemographic variables (Chandra et al., [8]; Fergus et al., [14]; Halpern & Haydon, [19]), all assessed at S1. Age was reported in years. Race or ethnicity was coded as three dichotomous indicators (Hispanic/Latino American, Black/African American, Asian American), with non-Hispanic/Latino, White/European American as the reference group. Mother's education, our proxy for SES, was measured using a 6-point scale from grade school or less (= 1) to graduate or professional school (= 6).

We ran a series of logistic multilevel growth curve models predicting the five sexual and two contraceptive behaviors. Models estimate coefficients that, when exponentiated, represent the odds of engaging in a particular sexual behavior. Sexual behavior models included all person-waves of data, and contraceptive behavior models included person-waves of data when individuals reported penetrative sex. For each behavior, we ran models in three steps. Model 1 examined odds of engaging in each behavior in the past three months as a function of time, indexed as S1 = 0 through S7 = 6, and included known correlates of sexual behavior: age, race or ethnicity, and mother's education. We tested linear, quadratic, and cubic terms for time; however, because the quadratic and cubic terms were not significant in any model, we report only linear terms for time. Model 2 examined differences in each behavior across time by the prior predictors and by gender and relationship status. The final step (Model 3) added a three-way interaction of time, gender, and relationship status. This final model was estimated using the following equation at Level 1 (within-person):

Sexti = β0i + β1i (Semester) + β2i (Serious Relationship) + β3i (Serious Relationship × Semester) + rit.

Here a sexual or contraceptive behavior at a given wave for a given individual is predicted by an intercept, the semester, whether the individual was in a serious relationship that semester, the interaction between semester and relationship status, and a within-person error term. The Level 2 (between-person) equations were the following:

β 0i = γ00 + γ01 (Age) + γ02 (Hispanic/Latino American) + γ03 (Black/African American) + γ04 (Asian American) + γ05 (Mother's Education) + γ06 (Male) + U0i

β 1i = γ10 + γ11 (Male)

β 2i = γ20 + γ21 (Male)

β 3i = γ30 + γ31 (Male)

The intercept β0i was predicted by the average odds for the reference group (White, non-multiracial female students of mean age at S1) plus terms for age in years, race or ethnicity, mother's education, gender, and an error term. The effect of semester consisted of the average effect for female students plus the difference for male students and an error term allowing for variability in the effect of semester over time. The effect of relationship status was predicted by the effect for female students plus the difference for male students, as was the gender difference in the semester by relationship status interaction.

Results are presented in Table 1. In Model 1, several control variables were significant. Hispanic/Latino American students were more likely than European American students to report performing oral sex in the past three months, but did not differ on the other six behaviors. When engaging in penetrative sex, African American students were less likely to report using any contraception consistently, but did not differ on the other six behaviors. Asian American students were less likely to engage in all of the behaviors in the past three months, with the exception of using condoms. Students whose mothers were more educated were more likely to both perform and receive oral sex, and to use any contraception and condoms.

Table 1. Multilevel Growth Curve Models Examining Trends In Past Three Month Sexual Behavior Over Seven Semesters of College and Differences by Gender and Relationship Status.

1 Note. *p < .05; **p < .01; ***p < .001. aReference category is non-Hispanic/Latino, White/European American.

In Model 1, semester was significant for all seven measures. Specifically, over time, the likelihood of students kissing, touching, performing oral sex, receiving oral sex, and engaging in penetrative sex increased (see Figure 1). For example, estimated rates of penetrative sex in the past three months increased from about 27% at S1 to 51% at Semester 7 (S7). The likelihood of using condoms, or any form of contraception, decreased (see Figure 1). For example, estimated rates of consistent condom use decreased over the course of the study from 57% to 29%.

Graph: Figure 1. Estimated rates of sexual and contraceptive behavior in the past three months across seven semesters of college. As indicated in Table 1, Model 1, semester was significant for all seven behaviors.

We next turn to Model 2, in which we added gender, relationship status, and their interactions with semester. Gender was significant for two of the seven measures. Specifically, men were more likely to report receiving oral sex and using condoms in the past three months than women were. Relationship status was significant for all five sexual behaviors. Students were more likely to report recently engaging in kissing, touching, performing oral sex, receiving oral sex, and penetrative sex when they were in a serious relationship compared to when they were not (see Figures 2–3). In addition, relationship status was significant for using condoms, but not for using any contraception. Students were less likely to report using condoms when they were in a serious relationship compared to when they were not. With the addition of gender and relationship status, semester remained significant in six of the seven analyses, but was no longer significant for receiving oral sex. Interactions between gender and semester or relationship status and semester were not significant for any of the sexual or contraceptive behaviors.

Graph: Figure 2. Estimated rates of kissing and touching in the past three months across seven semesters of college: differences by gender and serious relationship status. Relationship status varies across semesters, so, for instance, "Male, Serious Relationship" indicates male participants in semesters they were in a serious relationship. As indicated in Table 1, Model 2, relationship status (but not gender) was significant for both behaviors.

Graph: Figure 3. Estimated rates of oral and penetrative sex in the past three months across seven semesters of college: Differences by gender and relationship status. As indicated in Table 1, Model 2, relationship status was significant for all three behaviors; gender was only significant for receiving oral sex.

Finally, in Model 3 we added the three-way interaction among semester, gender, and relationship status. The three-way interaction was not significant for any of the five sexual behaviors, and therefore we do not interpret Model 3 for these behaviors. The three-way interaction was significant for both of the contraceptive behaviors. With the three-way interaction in the model, semester was no longer significant for any contraception and condom use. Results are plotted in Figure 4, and suggest that consistent use of any contraception in the past three months decreased over time more for men in semesters when they were not in a serious relationship than for men in semesters when they were in a serious relationship or for women. Figure 4 also illustrates how condom use decreased for men regardless of relationship status, and for women in semesters they were in a serious relationship, but did not decrease for women in semesters when they were not in relationships.

Graph: Figure 4. Estimated rates of any contraception use and condom use during penetrative sex in the past three months across seven semesters of college: Differences by gender and relationship status. As indicated in Table 1, Model 3, the three-way interaction among semester, gender, and relationship status was significant for both behaviors.

This study used longitudinal data to demonstrate that rates of recent sexual and contraceptive behaviors changed across college. Specifically, nonpenetrative and penetrative behaviors increased, and contraceptive behaviors decreased. Gender differences were limited: Men were more likely to receive oral sex and use condoms than were women. In contrast, in support of the contextual aspects of a normative developmental framework, relationship status mattered. Students were more likely to engage in all sexual behaviors, and less likely to use condoms, in semesters they were in serious relationships. For contraceptive behaviors, the importance of relationship status varied by gender.

Participation in nonpenetrative sexual behaviors was normative across all waves. For instance, at the start of college, most (80%) students had kissed in the past three months, and this rate rose to almost 90% by senior year, though this rate was lower for young adults who were not in relationships. Participation in touching was almost as prevalent and increased similarly. We believe these are the first published rates of recent kissing and touching in a traditionally aged college sample. Lifetime rates of kissing and touching are similarly high (O'Sullivan et al., [33]; Regan et al., [42]; Smiler et al., [49]), and our recent behavior rates suggest that the majority of students continue to engage in these behaviors somewhat regularly. Rates for penetrative and oral sex increased modestly as well, from about 20–30% at the start of college to about 25–40% by senior year. Although limited past work has examined longitudinal changes, our findings are in line with prior cross-sectional work that suggests similar age differences (Herbenick, 2010; Smiler et al., [49]; Williams et al., [59]). These results provide support for a developmental framework on sexuality that acknowledges the normative nature of sexual experience in this period. Engaging in nonpenetrative and penetrative sexual behaviors across adulthood is associated with positive physical, mental, and relationship health (Ebrahim et al., [13]; Lefkowitz et al., [27]; Nicolosi et al., [32]; Sprecher, [50]; Vasilenko et al., [55]). Given these known associations, it is possible that the increasing rates of engaging in sexual behaviors across college demonstrated here contribute to increased well-being during this developmental period. Future research might consider whether such health benefits are especially true for nonpenetrative behaviors such as kissing and touching, which do not have accompanying risks of unwanted physical health outcomes such as STIs or unplanned pregnancy.

In contrast, rates of consistently using any form of contraception did not increase with age, as may be expected from individuals growing older and gaining maturity and experience. Despite very few students being in married or cohabiting relationships, and very few likely wanting to get pregnant, in their first semester about 85%, and by senior year about 77%, reported always using a form of contraception when engaging in recent penetrative sex. Prior cross-sectional data on age differences in contraception use have been mixed (Manlove et al., [29]; Santelli et al., [46]; Whitaker et al., [58]; Wu et al., [60]). Our findings suggest some decrease in contraception use over time. However, these time trends were in part moderated by gender and relationship status, which may explain why such findings have been hard to demonstrate in prior work. Moreover, our question did not differentiate between more and less effective types of contraception. Past research suggests that a substantial proportion of college students who use contraception report less effective methods such as withdrawal or spermicide, and thus, even many students in our study who reported using contraception may not have been adequately protected from unwanted pregnancy (American College Health Association, [2]). Compared to contraception in general and despite significant risk for STIs, condom use rates started lower and decreased even more dramatically, from about 55% at the start of college to below 30% by the end. These findings demonstrate that although college students in their late teens to very early 20s frequently take some steps to protect themselves from unwanted pregnancy, potential exposure to STIs through sexual behaviors remains quite common. To prevent unwanted outcomes associated with sexual behavior in this developmental period, researchers and policy makers must acknowledge that sexual behavior is normative, while simultaneously recognizing inherent risks.

Men and women did not differ in their reported rates of kissing, touching, performing oral sex, penetrative sex, or using any form of contraception in the past three months. Although a number of prior studies have reported gender differences in such rates, particularly when reporting lifetime behaviors (Garcia et al., [17]; Herbenick et al., [22]; Petersen & Hyde, [39]), they may in part be due to intentional over- or underreporting or differences in estimation strategies (Alexander & Fisher, [1]) rather than true differences.

However, men were more likely to report recently receiving oral sex than women were. Unlike gender differences in other behaviors between men and women, gender differences in oral sex are more plausible, because the performing and receiving partners have different roles. This finding is important in understanding the experience of pleasure and orgasm, as it suggests that young men are more likely than women to receive direct genital stimulation through oral contact. Given that men are more likely than women to achieve orgasm from vaginal sex, and that women are more likely to achieve orgasm if sexual encounters include receiving oral sex (Richters, De Visser, Rissel, & Smith, [43]), the gender difference in oral sex further increases the gender gap in stimulation and potential orgasm. It also has implications for the normative experience of sexual behaviors during this developmental period if men are more likely to receive physical pleasure through both vaginal and oral sex than are women. Because women are more motivated than men to have sex for intimacy reasons and less motivated by physical pleasure (Cooper, Shapiro, & Powers, [11]), differences in oral stimulation could contribute to an increased gap between men's and women's perceptions of sexual experiences, perpetuating norms that men seek sexual experiences for physical satisfaction more than women do.

Across college and relationship types, men were more likely than women to report recent condom use, consistent with some prior research (Bailey et al., [4]; Petersen & Hyde, [39]; Pingel et al., [40]). This finding may be due to men being more likely to report condom use simply because they physically wear the condom, and thus it is more salient. On the other hand, it is also possible that this gender difference accurately captures age changes in condom use. That is, because women tend to have slightly older partners (Kaestle, Morisky, & Wiley, [24]), and rates of condom use decline over time, the gender difference may reflect this age-time trend.

Students were more likely to report all five sexual behaviors in semesters when they were in serious romantic relationships, consistent with past cross-sectional research (Fielder et al., [15]). This finding may reflect the availability of partners for students in relationships, and that sexual behavior is a key factor of romantic relationships (Collins et al., [10]). The unique contribution of our finding is that we used within-person analyses, thus demonstrating significant variations in sexual behaviors within the same individuals across time as a function of changes in relationship status. That is, the observed differences in rates of sexual behaviors by romantic relationship status were not simply due to stable between-person differences in the types of people who seek out, establish, and maintain romantic relationships.

Although rates of behaviors differed by relationship status, relationship status did not explain the time trend for the majority of sexual behaviors. As students progressed through college, their likelihood of recent kissing, touching, performing oral sex, and penetrative sex increased regardless of relationship status. Thus, these trends exist independent of romantic relationship experiences. Young men and women, then, may have sexual experiences regardless of relationship context, at a developmental stage when they are closer to establishing long-term relationships. This increase in sexual behavior outside of serious romantic relationships could alternatively represent simple non-goal-directed expressions of sexual interest, desire for sexual experiences before settling into committed relationships, or attempts to build relationships with potential long-term partners. Evidence suggests that romantic relationships that start as friends with benefits relationships are similar to relationships that do not, so this latter strategy might be effective for subsequently establishing serious romantic relationships (Owen & Fincham, [36]). Future work should consider not only rates of behaviors, but also motivations for sexual behavior outside of serious relationships, and whether these motivations change with time.

One important finding is that likelihood of engaging in penetrative sex differed greatly by relationship status, and was relatively low when students were not in a serious relationship—around 20% at the start of college and less than 40% by the end. These findings are consistent with cross-sectional research with the same sample, suggesting that the majority of nonrelationship sexual experiences do not involve penetrative sex (Wesche et al., [57]). Thus, despite the media attention given to college hookup culture (Fielder et al., [15]), the majority of penetrative sex in this college sample occurred within a serious romantic relationship.

The role of romantic relationships in consistently using any contraception was more complex. Prior cross-sectional research is mixed on associations between age and any contraception use (Manlove et al., [29]; Whitaker et al., [58]; Wu et al., [60]), but does suggest that individuals in romantic relationships are more likely to consistently use any contraception (Bailey et al., [4]; Ott et al., [35]; Pingel et al., [40]). In the current study, using any contraception decreased over time more for male students in semesters when they were not in a serious relationship, compared to other groups. This finding may be because individuals move from condoms to other forms of contraception as they get older. Because hormonal methods, which are only available for women, increase during this period (Santelli et al., [46]; Whitaker et al., [58]), women are more likely than men to know when hormonal contraception is used. Men in serious relationships, however, are more likely to have discussed contraception or STIs with their partners, thus knowing whether their female partner is using a hormonal method (Ryan, Franzetta, Manlove, & Holcombe, [45]). However, as women move toward hormonal methods, men having sex outside of serious relationships may be less knowledgeable about whether their sexual intercourse was protected by hormonal methods. Discussion of STDs and contraception between hookup partners is relatively infrequent (Downing-Matibag & Geisinger, [12]). Advocates for increased equity, communication, and affirmative indications of consent might also work for men to share more responsibility for active participation in contraception use.

The findings for condom use were slightly different. We replicated prior cross-sectional work (Bailey et al., [4]; Petersen & Hyde, [39]) that individuals in serious relationships reported less frequent condom use. However, accounting for time trends revealed that recent condom use decreased for men regardless of relationship status, and for women in semesters they were in serious relationships. Condom use did not decrease for women in semesters they were not in serious relationships. The finding for women is expected, because women in relationships are more likely to switch to hormonal birth control (Civic, [9]). Men and women may have different opinions of condoms, given that the majority of young men believe that condoms will reduce their pleasure, and these attitudes are associated with lower condom use (Higgins & Wang, [23]). One possible explanation, then, is that experience with condom use during sex may decrease men's interest in using condoms, whether with serious relationship partners or not.

This study had limitations that warrant caution and suggest future directions. First, the sample was limited to college students at one large residential state university in a college town. The contexts and opportunities underlying sexual relationships, and thus rates of sexual and contraceptive behaviors, likely differ for college students in more urban areas, or at religious, nonresidential, or online universities. In addition, experiences for young adults who do not attend college likely differ, in part because their living arrangements afford less access to same-age peers. Nonstudents may meet older or younger partners, which could alter their behavioral options and patterns, particularly when it comes to risk behaviors. Although we did not exclude sexual minority young adults, this sample was predominantly heterosexual, and thus, we could not examine differences by sexual identity. Similarly, although the sample was ethnically and racially diverse, we did not consider differential changes by ethnicity or race. Future work should consider all of these other populations to better understand sexual behaviors in a wider range of individuals and contexts. Second, we assessed recent sexual behaviors (prior three months), but current romantic relationship status. Given the volatility of romantic relationship status in this age period (Halpern-Meekin, Manning, Giordano, & Longmore, [20]), it is possible that someone who was single at the time of assessment had engaged in sexual behaviors with a romantic relationship partner two months prior. Future work on recent sexual behaviors should match the time frame more tightly with romantic relationship status. Third, our question asked about penetrative sex, combining vaginal and anal sex. Future work should consider these behaviors separately. Fourth, we did not ask whether the sexual behavior was consensual, and some of the behaviors we assessed were likely nonconsensual and thus not the young adults' choice (Mellins et al., [30]). Finally, we focused on sexual behaviors across the college years, and experiences before and after college likely differ. Future work should consider longer-term developmental trends beginning in early adolescence, such as modeling the initiation, exploration, and development of a variety of sexual and contraceptive behaviors. In addition, as individuals leave college to new work and relationship roles, understanding changes in sexual and contraceptive behaviors within these relationships will be important.

Despite these limitations, this study demonstrates that across college, rates of recent participation in penetrative and nonpenetrative sexual behaviors generally increase, whether young adults are in serious romantic relationships or not. Still, romantic relationship status was an important determinant of engaging in all types of sexual behaviors, as well as consistently using any contraception. Findings suggest that reporting using any contraception decreased over time more for men in semesters when they were not in serious relationships than for men in semesters when they were in relationships or for women. Many universities include sexual health promotion programming during first year orientation. Our findings suggest the value of continued sexual health promotion, with a particular focus on contraceptive behaviors, across the college years.

The authors would like to thank the University Life Study team for their help with data collection and data management.