The Prevalence, Frequency, and Social Ecology of Sexual Concurrency Among Young Adult Women

Abstract

CONTEXT:

Sexual concurrency is associated with increased risks of STI transmission and unintended pregnancy. Better understanding its prevalence and social ecological correlates among women is imperative to understanding the proliferation of sexual health disparities.

METHODS:

Weekly population-representative panel data from 757 women aged 18 to 22 years, collected in one Michigan county from 2008–2012, were drawn from the Relationship Dynamics and Social Life study. Univariate analyses assessed the prevalence of two forms of concurrency across women and their frequencies within women over time. Multivariate logistic regressions investigated which characteristics of women’s social ecologies predicted concurrency in a given week.

RESULTS:

Twenty percent of women had intercourse with two partners in one week; 14% had intercourse with a second partner in the middle of an ongoing relationship. In both cases, the majority had intercourse with the second partner in 1–3 weeks total. Both types were most common among women who believed they should have sex with men after seeing them a while; were less willing to refuse sex; African American; and in nonexclusive relationships. Sex with multiple partners in one week was also most common among women whose parents disapproved of and least educated them about sex; whose friends approved of sex; and who had sex without contraception before baseline.

CONCLUSIONS:

Sexual concurrency among young women is prevalent but intermittent. Young women’s likelihood of becoming concurrent varies depending on complex social ecological factors including their sexual beliefs, norms, and history; relationship type; and race.

Sexual concurrency—having sexual partnerships that overlap in time—is associated with an elevated risk of sexually transmitted infections,^[1–5] in large part because it connects healthy individuals to potentially infected ones through a mutual partner.^{[6, 7]} When someone has sex with multiple partners in quick succession it further compounds the risk of STI transmission by reducing the amount of time that person has between partners to become symptomatic and to get tested or treated.^[8] Moreover, individuals who engage in concurrency are more likely than others to also engage in a range of risk-taking behaviors, including unprotected sex and substance use, that also increase the risk of STI transmission and unintended pregnancy.^{[9, 10]} Understanding who is most likely to be concurrent and under what circumstances they are most likely to do so is thus central to better understanding the etiology of STIs and the perpetuation of sexual health disparities.

Considering that incidence rates of STIs and unintended pregnancy are highest during the transition to adulthood,^[11–13] expanding our knowledge of concurrency among young women is especially important. Few studies have examined concurrency among women specifically, and among those that have, most aggregate women from a wide age range.^{[3, 10, 14–16]} At least two studies, however, indicate that concurrency is more common at young ages,^{[10, 17]} thus highlighting the need for a greater emphasis on concurrency among young adults. Moreover, because past scholarship has largely relied on cross-sectional data,^{[10, 15, 18]} scholars have rarely been able to identify instances in which intercourse with both partners occurred in a narrow timeframe or examined how the likelihood of concurrency varies alongside women’s changing social ecology, meaning their ideation, or developing thoughts, and their multi-dimensional social environment.

The present study advances existing research by using panel data collected at short, one-week intervals to estimate the prevalence and frequency of concurrency among young adult women and to investigate its relationship to women’s dynamic social ecological characteristics. Using repeated, weekly observations on sexual activity and relationships, we asses and compare two types of concurrency: (1) short, one-week windows in which women had vaginal intercourse with two or more partners; and (2) instances in which women were in an ongoing heterosexual relationship and had vaginal intercourse with at least one other person in the middle of that relationship.

MEASUREMENT OF SEXUAL CONCURRENCY

With a few exceptions,^{[9, 19]} most research on concurrency in the United States has relied on cross-sectional data in which concurrency is assessed by asking respondents to identify the first and last time they had intercourse with each of their recent partners;^{[10, 20–22]} to report additional people they had sex with while in their most recent relationship;^[21–24] and/or whether they are presently in multiple, ongoing relationships.^[20] None of these approaches necessarily identifies how close in time a person had sex with each partner—an important piece of information that affects the risk of STI transmission^[8]. Moreover, the former two approaches suffer from underreporting because they rely on respondents’ memory.^{[21, 22, 25]}

Relying on cross-sectional data or data collected at infrequent intervals^{[18, 26, 27]} also prevents scholars from estimating how the likelihood of concurrency varies with the ebb and flow of various facets of women’s lives. Yet, like young women’s sexual desires, attitudes, and agency,^[28–30] women’s sexual behaviors^[31]—including concurrency—may change as they progress through romantic relationships, acquire new friends, or enter new educational settings.

THEORETICAL FRAMEWORK

Notwithstanding common data limitations, extant scholarship indicates that between approximately 6% and 8% of reproductive-aged women in the United States have concurrent relationships.^[10] However, given the density of sexual experimentation that occurs during adolescence and early adulthood,^[32] concurrency may be more common during adolescence and the transition to adulthood than at later stages of the life course.^{[10, 17]}

Relatedly, women more often report having concurrent partners when they are in in non-exclusive or non-married relationships relative to exclusive or married ones.^{[10, 24]} One study, however, finds that among young women, approximately half of concurrent incidents (where women have sex with more than one person in one week’s time) occur within exclusive relationships.^[9] This may be in part because women are more likely to have sex with secondary partners when they mistrust their primary partner or believe that he has recently had sex with someone else.^{[22, 33]}

Studies further suggest that concurrency is more prevalent among women who sexually debuted at an early age (during or before early adolescence),^{[10, 34]} who were abused as children,^[17] and who have previously been sexually coerced.^[17] Variation across past sexual experiences and abuse indicates that concurrency may be more common among women who lack (a sense of) sexual agency. This latter interpretation is supported by qualitative research that finds that some women who would prefer to be in exclusive relationships stay with partners who are non-monogamous^{[35, 36]} and that, in some instances, they blame other women for their partners’ extra-relationship sexual activity.^[35]

Patterned variation in the prevalence of concurrency across sexual histories and pivotal earlier life experiences highlight the socially embedded nature of concurrency. According to the social ecological model of human development,^[37] individuals’ thoughts, feelings, and behavior share a reciprocal relationship with their social surroundings. Conceptually, this model views women’s social ecology as consisting of five mutually reinforcing layers: women themselves; their social networks; the organizational settings that facilitate network interactions; the broader community that these organizations belong to; and the encompassing social structural and policy environment. Drawing on a social ecological framework suggests that young women’s likelihood of being concurrent should be related to their personal perceptions of sex; the people whom they interact with the most, e.g. partners, friends, and family; sexual norms among those individuals; and the institutional settings they belong to like school and work.

Studies of related sexual outcomes provide support for this perspective. In particular, they highlight connections between young women’s sexual attitudes and behavior and those of their family and friends, partners, and school and religious environments. For instance, one study finds that the more comfortable adolescents are discussing sex with their parents, friends, and partners, the more they intend to delay sexual activity, and correspondingly, the less likely they are to be sexually active.^[38] Likewise, another study finds that young women are less likely to have sex in adolescence when their parents speak with them about abstinence.^[39]

Studies of peer groups in particular find that women’s condom use and risk of early pregnancy tend to mimic the norms among their friends and within their school settings.^{[31, 40, 41]} In fact, one study finds that how much young women want to have sex depends on how many of their friends are sexually active at that time.^[28] Partners, too, play an important role in the sex lives of young women. Studies show that women’s sexual activity corresponds to the characteristics of their partners, like whether he is older and more sexually experienced than her.^[42]

Moreover, studies documenting racial and socioeconomic differences in women’s sexual and reproductive attitudes find that these differences are often attributable to distinct social norms, neighborhood contexts, and institutional memberships, as indexed by differences in religiosity and college enrollment.^{[28, 43–45]} Nevertheless, numerous studies find that, net of other factors, race is a strong predictor of concurrency, with African American women more likely to report having concurrent relationships than women from other racial/ethnic backgrounds.^{[10, 15]} Considering the infrequency of interracial relationships,^[46] and thus the racial segregation of sexual networks, higher rates of concurrency among African Americans may help to explain a higher prevalence of STIs among African Americans than among other groups.^[47]

Of course, social ecologies are not static but rather evolve as a function of pivotal experiences that change individuals’ thoughts and preferences over time.^[48] For instance, early sexual experiences may be related to subsequent concurrency by influencing young women’s ideas about what is “normal” or acceptable, their social mores, and the perceived consequences they associate with different decisions. Drawing on the social ecological model, the limited body of research on concurrency among women, and research on women’s sexual and reproductive health more broadly lead us to expect that young women’s likelihood of having concurrent partnerships should vary with their sexual ideation (evolving attitudes and perceived agency), perceived norms among family and friends, previous sexual experiences, and select demographic characteristics like race and relationship commitment.

METHODS

Sample and Data Collection

We use data from the Relationship Dynamics and Social Life Study (RDSL), a population-representative weekly survey of young women who were 18 or 19 years old and residing in one Michigan county at baseline. Women who were temporarily residing outside the county at the time of recruitment were also included. Participants were randomly selected from the Michigan Department of State driver’s license and Personal Identification Card database, with an 84% response rate (94% of successfully located individuals).

Baseline data were collected in-person on a rolling basis between 2008 and 2009. Respondents were given $20 for taking the baseline survey, which assessed a wide range of psychosocial and behavioral factors related to sexual and reproductive health. These included respondents’ sociodemographic background; sexual and reproductive history; and reproductive attitudes, agency, and perceived norms.

At the close of baseline, respondents were invited to participate in the weekly, journal-portion of the study, which took about 5 minutes to complete online or over the phone. These journals asked respondents about their relationship status, sexual activity, and contraceptive use. Every twelve weeks they also reassessed respondents’ time-varying demographics (like school enrollment) and their sexual ideation and perceived norms. Respondents were paid $1 for every completed journal and $5 for every five journals consecutively completed on time. Seventy-eight percent of respondents completed journals for at least 1.5 years, while 63% completed journals for 2.5 years.^[49] An experiment conducted in tandem with the RDSL confirmed that repeatedly completing journals had little influence over respondents’ attitudes or behavior.^[50]

Missing data were rare, with only 3% of item-specific data missing.^[49] Moreover, missing data in weekly journals appeared to be at random—women typically did not skip multiple questions in the same week nor were the same women repeatedly missing responses to the same questions across weeks. To retain weeks with missing information on time-varying predictors, we therefore replace missing values with the value a woman most recently reported. Re-estimating our models using listwise deletion yields substantively similar results (available upon request).

Because of this study’s emphasis on concurrency, we limit all analyses to women who were ever sexually active during the study and who completed two or more journals. From these women we create two analytic samples. Considering that a woman could engage in both types of concurrency, these samples are not mutually exclusive. The first is comprised of weeks (journals) when respondents reported having sex (N=19,669 weeks from 757 women). This allows us to assess the predictors of whether a woman had sex with more than one person, versus with one person only, in a given week. The second analytic sample is comprised of weeks when respondents were in a sexually active relationship, and more specifically, were in the middle of that relationship (e.g. not in its first or final week, N=21,244 weeks from 638 women). This enables us to assess what predicts whether a woman had sex with a secondary partner after she had already begun having sex with a primary partner and before she stopped having sex with him (what some refer to as “embedded” concurrency).^[51]

Measures

• Sexual concurrency.

Each week in which respondents reported being in a relationship they were asked to provide their primary partner’s initials. They were then asked whether they had vaginal intercourse with that partner that week. This allows us to identify all primary relationships that were sexually active and to specifically identify when they first and last had sex. In weeks when respondents reported being in a relationship, they were also asked whether they had vaginal intercourse “with anyone other than [primary partner’s initials].” This additional information enables to create two measures of concurrency. The first indicates whether a woman had sex with two or more people in a given week, meaning whether she reported intercourse with her primary partner and with somebody else that week, versus intercourse with one person only (either her primary partner or somebody else). The second indicates embedded concurrency—when a woman reported that she had intercourse with anyone other than her primary partner, versus not, specifically in a week that fell between the first and last week she had sex with her primary partner.

• Sexual ideation.

Sexual ideation consists of respondents’ sexual attitudes, self-efficacy, and perceived agency, which were updated every twelve weeks. Attitudes were assessed by asking respondents whether they believed that a woman “who has been seeing a guy for a while should have sex with him,” with possible responses ranging from (1) strongly disagree to (5) strongly agree. Self-efficacy was assessed by asking respondents what the chances are, on a scale of 0 to 100, that they would be able to stop themselves from having sex once aroused. For the ease of interpretation, we divide original responses by 10. Agency was measured by asking respondents how willing they would be to refuse unwanted sex even if it made their partner angry, with responses ranging from (0) not at all to (5) extremely.

• Norms.

The assessment of associations between concurrency and social norms includes four indicators. The first is how much informal sex education respondents received from their parents. This is measured with an index, ranging from 0 to 5, that sums affirmative responses to questions about whether a respondents’ parents ever spoke to her about how to say no to sex, methods of birth control, where to get birth control, sexually transmitted diseases, and how to use condoms. The second and third are how positively her parents and friends (separately) would react if they found out she was having sex, ranging from (0) not at all to (5) extremely. The fourth is the proportion of respondents’ friends who were sexually active, from (1) none to (5) almost all. Perceptions of approval and of sexual activity among friends were reassessed every twelve weeks.

• Institutional membership.

We consider two types of institutional membership—education and religion. Education is defined as whether a respondent was currently attending a 2- or 4-year college, which was updated every twelve weeks. Because 95% of respondents reported belonging to a Christian denomination or having no religion, we conceptualize religion as whether respondents reported being highly religious, meaning whether they reported that their religion was either “very important” or “more important than anything else” versus “somewhat” or “not important.” Religiosity was assessed only at baseline.

• Risky sexual behavior.

History of risky sexual behavior includes markers associated with early pregnancy and STI transmission, namely whether a respondent first had sexual intercourse at age 14 or younger and whether she had ever had sexual intercourse without contraception by baseline.

• Respondent and partner demographics.

We assess five demographic characteristics of respondents. These include her parity (cumulative number of births to date, including those that occurred before baseline); age; whether her mother graduated college; whether she grew up in a two-parent home; and her race, which is defined as whether she self-identified as African American or not. This focus on African American identity reflects that 97% of respondents identified as either African American or white. Partners’ demographics, which were reported by respondents about their primary partners only, include whether they had ever attended college and their age. We do not control for partners’ race because this is highly correlated with respondents’ race. Finally, we also control for the respondent and her partner’s current relationship status, defined as nonexclusive, exclusive, cohabiting, and married or engaged.

Analysis

To begin, we calculate the percent of sexually active respondents that ever had sex with two partners in the same week and that ever had sex with a second partner while in an ongoing sexual relationship during the study period, separately. Then, among these women, we tabulate how many weeks in total they were concurrent (according to the corresponding definition).

Following, we assess the social ecological correlates of concurrency using logistic regressions with random effects. Random effects models address correlated errors within respondents by assigning each respondent a specific coefficient and, at the same time, address correlated errors within weeks by assigning each week a coefficient. This means that coefficients on time-invariant indicators, like race, convey average differences in the log-odds of concurrency across respondents over the course of the study. Coefficients on time-varying indicators like education, however, are simultaneously derived from the estimated difference in the log-odds of concurrency across respondents with and without that trait (e.g. between those who were and were not enrolled in college in a given week) and the estimated difference in the log-odds of concurrency within respondents as they moved into or out of possessing that trait (e.g. between when the same woman was and was not enrolled in college). First we estimate a model predicting a woman’s odds of having sex with two or more partners in a given week, versus with one partner only. Following, we estimate an identical model predicting a woman’s odds of having sex with a secondary partner in weeks when she was amidst an ongoing, primary sexual relationship, versus not.

Because the RDSL consists of a simple random sample no survey weights are needed or applied. All analyses are conducted in Stata version 15.

RESULTS

Sample Characteristics

Sexual ideation, norms, institutional environments, sexual histories and demographic backgrounds are highly similar across both analytic samples (Table 1). This reassuringly suggests that any differences in how these measures predict the two forms of concurrency in the multivariate analysis should not be driven by compositional differences. For the sake of parsimony, we limit our discussion of descriptive statistics to the sample used to assess the likelihood of having two or more partners (in the left-hand columns of Table 1). When discussing time-variant characteristics, we describe the sample of weeks; when discussing time-invariant characteristics, we refer to the sample of women.

Table 1.

Selected characteristics of respondents aged 18–22 during the Relationship Dynamics and Social Life Study, 2008–2012

	Mean or %		Mean or %
	Sample 1: Two partners in same week		Sample 2: Second partner during ongoing sexual relationship
	Weeks (N=19,669)	Women (N=757)	Weeks (N=21,244)	Women (N=638)
Sexual ideation +
Women should have sex with men after seeing them a while (1–5)	1.87 (.68)	1.88 (.60)	1.84 (.66)	1.86 (.59)
Perceived ability to stop herself once aroused (0–10)	6.90 (3.02)	7.68 (2.81)	7.02 (2.97)	7.69 (2.78)
Willingness to refuse unwanted sex (0–5)	3.81 (1.57)	4.05 (1.61)	3.82 (1.54)	4.12 (1.54)
Norms +
Informal parental sex education (0–5)	3.01 (1.16)	2.99 (1.79)	3.01 (1.72)	3.02 (1.77)
Parents’ approval of sex (0–5)	2.21 (1.27)	1.70 (1.48)	2.23 (1.26)	1.66 (1.45)
Friends’ approval of sex (0–5)	3.27 (1.16)	2.86 (1.41)	3.27 (1.13)	2.86 (1.40)
Proportion of friends sexually active (1–5)	4.52 (.84)	4.36 (1.00)	4.48 (.86)	4.38 (.99)
Institutional membership
Enrolled in college+
Yes	55	48	57	49
No	45	52	43	51
Highly religious
Yes	50	55	48	55
No	50	45	52	45
Risky sexual behavior
Sexual debut <=14 years old
Yes	20	19	18	19
No	80	81	82	81
Had sex without contraception before baseline
Yes	54	55	51	55
No	46	45	49	45
Respondent and partner demographics
Respondent
Parity+	.22 (.50)	.18 (.45)	.22 (.50)	.18 (.46)
Age+	20.30 (.93)	19.18 (.57)	20.35 (.92)	19.18 (.57)
Mother graduated college
Yes	23	20	24	22
No	77	80	76	78
Grew up with two parents
Yes	58	52	61	53
No	42	48	39	47
African American
Yes	22	34	19	31
No	78	66	81	69
Partner
>= Some college education
Yes	49		51
No	51		49
Age	22.80 (4.05)		22.77 (3.96)
Relationship status
Non-exclusive	9		4
Exclusive	46		48
Cohabiting	21		22
Married or engaged	24		26

⁺When describing samples of women, descriptive statistics pertain to baseline measurement.

In the average week, respondents did not have strong beliefs that women should have sex with male partners after seeing them a while—a score of 1.9 out of 5 (Table 1). Nevertheless, respondents’ perceived ability to stop themselves from having sex once aroused and their perceived willingness to stop their partners from having sex with them when they didn’t want to were only moderate. On a scale of 0 to 10, in the average week respondents reported their ability to stop themselves as 6.9. Likewise, they reported their willingness to refuse unwanted sex as 3.8 on a scale of 0 to 5.

Respondents similarly reported receiving moderate levels of informal sex education from their parents, with parents discussing an average of 3 out of 5 topics (Table 1). In a typical week, they expected their parents to react only somewhat positively, and their friends to react somewhat more positively if they found out that she was having sex, with scores of 2.2 and 3.3 respectively (on scales of 0 to 5). Despite only perceiving moderate approval from parents and friends, women typically perceived the majority of their friends to be sexually active, reporting an average of 4.5, or between “most” and “almost all” as sexually active in a given week (Table 1).

With respect to their institutional memberships, in 55% of weeks, respondents were enrolled in either a two-year or four-year college. More than half of respondents—55%—were highly religious (Table 1). In terms of their history of risky sexual behavior at baseline, 19% of respondents sexually debuted by age 14; 55% ever had sex without contraception by baseline (Table 1).

The two analytic samples are also highly similar in terms of respondents’ and their partners’ demographic characteristics. In the average week respondents had .22 children and were approximately 20 years and 4 months old (Table 1). One-fifth of respondents’ mothers graduated college, while slightly more than half—52%— grew up in a two-parent household (Table 1). Approximately a third (34%) of respondents were African American. In about half of weeks (49%), respondents’ partners had at least some college education (Table 1). On average, partners were approximately two and a half years older than respondents, with a mean age of 22 years and 10 months.

In both analytic samples, the most common type of relationship was exclusive (46% of weeks in the first and 48% of weeks in the second sample, Table 1). The two samples differed, however, with respect to non-exclusive relationships, which were more than twice as common in the sample assessing sex with two partners in the same week as in the sample assessing sex with a secondary partner in the middle of an ongoing primary sexual relationship (9% versus 4% of weeks, Table 1). On the flip side, cohabiting and married or engaged relationships were represented slightly more often (1 and 2 percentage points, respectively) in the latter sample than in the former (Table 1). These cross-sample differences in relationship status reflect that the second analytic sample is limited to weeks when women were in the middle of relationships that were sexually active for three weeks or more. More serious relationships were more likely to meet this criterion.

Descriptive Findings

Twenty percent of sexually active women (n=150, of 757 respondents) ever reported having intercourse with two or more partners in the same week (univariate analysis not shown).^a Of these women, 49% had intercourse with two or more partners in one week only; 25% in two weeks; and 11% in three weeks (Figure 1). Concurrency of this type was thus prevalent among women but infrequent across time.

Figure 1.

Frequency of Concurrency Among Ever-Concurrent Women

Fourteen percent of sexually active women who were ever in a relationship that was three weeks or longer (n=87, of 638 women) reported having intercourse with a non-primary partner while in an ongoing sexual relationship.^b Among these 87 women, 62% had intercourse with a non-primary partner in the middle of their sexually active primary relationship in just one week; 17% in two weeks; and 7% in three weeks (Figure 1). Embedded concurrency was thus also prevalent but infrequent.

There was substantial overlap between the two types of concurrency: In 40% of weeks when women had sex with two or more partners they did so in the middle of an ongoing primary sexual relationship. In the remaining 60% of weeks when women had sex with two or more partners it did not occur in the middle of an ongoing sexual relationship, either because their primary relationship was brief (<3 weeks) and/or because they had sex with a second partner in either the first or last week of their primary relationship.^[9] In 76% of weeks when women had sex with a secondary partner amidst their other ongoing sexual relationship they had sex with both their secondary and their primary partner in the same week (analysis not shown).

Multivariate Findings

Starting with sexual ideation, for each additional point that a woman more strongly believed that women should have sex with men after seeing them a while her log-odds of having sex with two or more people increased by .3, and similarly, her log-odds of having sex with a second partner during the middle of her primary sexual relationship increased by .2 (Table 2). Women’s odds of both forms of concurrency did not vary with their perceived ability to stop themselves from having sex once aroused. However, for each additional point a woman thought she would be willing to refuse unwanted sex, even if it made her partner angry, her log-odds of having sex with >=2 partners that week (versus with one person only) decreased by .1. In other words, the odds of having sex with multiple people in the same week are estimated to vary by as much as .6 between weeks when women were not at all willing to refuse unwanted sex (a value of 0) and extremely willing to refuse it (a value of 5). Likewise, for each additional point a woman thought she would be willing to refuse unwanted sex her log-odds of having sex with a secondary partner in the middle of her primary relationship (versus not) decreased by .1.

Table 2.

Results of logistic regression with random effects, expressed as changes in the log-odds (and 95% confidence intervals), assessing associations between sexual concurrency and social ecology

	M1: Two partners in the same week (N=19,669 weeks from 757 women)	M2: Second partner during ongoing sexual relationship (N=21,244 weeks from 638 women)
Sexual ideation
Women should have sex with men after seeing them a while	.27 (.08 – .45)***	.23 (−.00 – .46)*
Perceived ability to stop herself once aroused	.03 (−.02 – .07)	.02 (−.05 – .10)
Willingness to refuse unwanted sex	−.10 (−.20 – −.01)**	−.13 (−.28 – .02)*
Norms
Informal parental sex education	−.15 (−.26 – −.04)***	−.09 (−.24 – .06)
Parents’ approval of sex	−.10 (−.22 – .02)*	−.06 (−.22 – .10)
Friends’ approval of sex	.13 (−.01 – .27)*	.16 (−.04 – .37)
Proportion of friends sexually active	−.14 (−.35 – .08)	.09 (−.21 – .40)
Institutional membership
Enrolled in college	.17 (−.17 – .51)	.35 (−.11 – .80)
Highly religious	−.18 (−.63 – .27)	.07 (−.53 – .67)
Risky sexual behavior
Sexual debut <=14 years old	−.08 (−.63 – .47)	.09 (−.69 – .86)
Had sex without contraception before baseline	.82 (.35 – 1.29)***	.47 (−.19 – 1.13)
Respondent and partner demographics
Respondent
Parity	−.13 (−.51 – .25)	−.18 (−.71 – .36)
Age	.01 (−.21 – .24)	−.10 (−.45 – .26)
Mother graduated college	−.33 (−.91 – .24)	−.05 (−.88 – .78)
Grew up with two parents	−.14 (−.59 – .30)	−.28 (−.93 – .37)
African American	.58 (.05 – 1.11)**	1.02 (.32 – 1.71)***
Partner
>= Some college education	−.07 (−.52 – .37)	−.12 (−.74 – .50)
Age	.02 (−.04 – .07)	.06 (−.01 – .12)*
Relationship status (ref: non-exclusive)
Exclusive	−1.87 (−2.22 – −1.51)***	−1.82 (−2.52 – −1.12)***
Cohabiting	−2.15 (−2.72 – −1.58)***	−1.82 (−2.58 – −1.07)***
Married or engaged	−2.64 (−3.28 – −1.99)***	−2.21 (−3.22 – −1.20)***
Constant	−3.99 (−8.65 – .67)*	−5.08 (−12.14 – 1.98)

^*p<.05,

^**p<.01,

^***p<.001.

In terms of women’s perceived social norms, informal parental sex education and perceived parental approval of sexual activity were both negatively associated with having sex with two or more people in the same week. More specifically, for each additional topic a woman’s parents discussed with her, her log-odds of having sex with multiple partners in a given week decreased by .2, while for each additional increase in parental approval of sex, her log-odds of this type of concurrency declined by .1 (Table 2). This pattern of coefficients suggests that women who grew up in approving, sex-educational family settings are less likely to have sex with two or more partners in a very short window of time. However, neither parental sex education nor parents’ approval of sex were predictive of women having sex with a second partner during another ongoing sexual relationship (Table 2). In contrast to parents, for each one-unit increase in perceived approval of sex among friends, women’s odds of having sex with two or more people in the same week increased by .1. Perceived approval of sex among friends was not associated with a woman’s log-odds of having sex with an outside partner during her ongoing primary sexual relationship. The perceived proportion of women’s friends who were sexually active was not significantly associated with either form of concurrency (Table 2).

With respect to women’s institutional memberships, women’s odds of concurrency, no matter how it was defined, did not differ by college enrollment. Nor did women’s odds of concurrency differ between women who were and were not highly religious (Table 2). Turning to risky sexual behavior, women were no more or less likely to engage in either form of concurrency conditional on whether their sexual debut occurred at age 14 or younger. Women who had sex without contraception before baseline, however, had .8 higher log-odds of having sex with two or more partners in the same week than women who had not (Table 2).

Irrespective of how concurrency was measured, few differences in the odds of it occurring were detected according to respondents’ demographic characteristics. However, compared to their non-African American peers, African American women had .6 and 1.0 higher log-odds of having sex with multiple people in a given week and of having sex with a secondary partner amidst another ongoing relationship, respectively. Few differences in concurrency were also observed across partners’ demographic characteristics, with the exception that primary partners’ age was positively associated with a woman’s log-odds of having sex with a secondary partner while still in that primary relationship (Table 2). Women’s log odds of engaging in either type of concurrency were highest when they were in non-exclusive relationships and monotonically decreased with relationship commitment level.

DISCUSSION

Sexual concurrency is highly associated with unwanted sexual and reproductive health outcomes and plays an important role in STI transmission dynamics. Our analyses of population representative data indicated that, during the transition to adulthood, 1 in 5 sexually active young women has sex with two or more people in the same week, while roughly 1 in 7 have sex with someone other than their primary partner in the middle of an ongoing relationship. High typological overlap between the two types of concurrency indicated that when young women have sexually concurrent partners there is oftentimes an insufficient window for them to present STI symptoms or get tested. Nevertheless, when women are concurrent, no matter how it is defined, they tend to have sex with a second partner for only one to three weeks total.

Both types of concurrency are clearly connected to women’s sexual ideation. Women who view sex as non-obligatory within relationships and who feel that they themselves would be willing to refuse unwanted sex have lower odds of concurrency than others. While previous qualitative research suggests that some women tolerate their partners being concurrent even when they don’t want them to be,^{[35, 36]} suggesting that men’s concurrency is associated with women’s lack of agency within relationships, the findings presented here indicate that women’s lack of sexual agency and perceptions that women are not entitled to such agency is also positively associated with their own concurrency.

Relatedly, when defined as sex with two or more people in the same week, concurrency is associated with sexual norms within women’s families—a key source of education and socialization during childhood and adolescence. Women who receive more sex education from their parents, including about topics like how to refuse unwanted sex, and women whose parents are more approving of sex are less likely to have sex with multiple people in the same week. These women are no more or less likely, however, to have sex with a secondary partner while in an ongoing relationship than their peers. It may therefore be the case that parents’ communication about sex contributes to young adults’ perceptions of the risks associated with different sexual scenarios, leading them to be more cautious of having sex with multiple people in quick succession but not necessarily of having sex with secondary partners over a longer window of time.

In contrast to parents’ approval, friends’ approval of sex is associated with a higher likelihood that women will have sex with more than one person in a one-week window. This juxtaposition highlights the value of taking a social ecological approach to understanding sexual concurrency during the transition to adulthood—an approach that recognizes the complexity of young people’s social lives, which at this developmental stage often includes weakening ties to family members and strengthening ties to peers.

Notably, neither form of concurrency is associated with a young woman’s institutional memberships, namely their college enrollment or religiosity. Research on “hook-ups” suggests that casual sexual encounters are quite common among college students.^{[52, 53]} Although sex with secondary partners is not necessarily casual, this study’s findings suggest that having sex with two or more people in the same week is no more or less common among college students than their non-college-going peers. Moreover, one of the largest predictors of having sex with a secondary partner during an ongoing relationship is a young woman’s earlier history of risky sexual behavior, namely whether she had sex without contraception. Thus, as others have shown,^[9] sexual concurrency is part of a broader pattern] of sexual risk-taking that is often established earlier in the life course.^{[10, 34]}

Neither form of concurrency differs much along demographic lines, with the exception of race. Race is the second largest predictor of whether a woman will have sex with a secondary partner amidst another ongoing relationship and the third largest predictor of whether she will have sex with two or more partners in the same week. The observed magnitudes of racial differences in concurrency dovetail with past research, which finds that concurrency is more prevalent among African Americans than among other racial and ethnic groups in the United States.^{[10, 15]} Also consistent with past studies,^{[10, 24]} relationship status is the largest predictor of both types of concurrency, highlighting the particularly protective role of exclusive, cohabiting, and married or engaged relationships.

Limitations

Several limitations are worth noting. First, although RDSL data are representative of a county that falls close to the national median on many demographic measures^[54] they are not nationally representative. Nevertheless, previous studies comparing the RDSL to nationally representative samples of women of the same age find that, with respect to related reproductive behaviors like nonmarital and teen childbearing, the RDSL is similar.^{[55, 56]} Second, the RDSL sample is limited to women aged 18 to 22. We are therefore unable to compare the prevalence, frequency, and social ecological correlates of young adult women to those of young adult men or to those of adolescent or older women. Third, our estimates are based on data that were collected approximately ten years ago. Young women’s behavior and ideation may have shifted over the interim.

Fourth, we are only able to identify instances of sexual concurrency that involved vaginal intercourse with two partners. This means we overlook concurrency when it involves only anal or oral sex. Fifth, and relatedly, although we identify when women had intercourse with two or more people in the same week, we do not know whether her relationship with the secondary partner was ongoing (and potentially involving other sex acts). Finally, because we do not have information on drug and alcohol use, we are unable to assess how concurrency varies with substance use—an important relationship that others have documented.^{[21, 22, 24]} To the extent that substance use is correlated with women’s sexual ideation, norms, and early experiences, its omission may upwardly bias our multivariate point estimates.

Conclusions

Sexual concurrency is fairly common among women during the transition to adulthood, yet most women who are ever concurrent are only concurrent intermittently. Sexual attitudes and perceived sexual agency are significantly associated with the likelihood that a young woman will have sex with two or more partners in the same week or have sex with a partner outside her ongoing primary relationship. Other sexual facets of women’s social ecology, like their parents’ approval of and communication about sex, friends’ approval of sex, and personal experiences with unprotected sex at any early age are also associated with their likelihood of having sex with two or more partners in a one-week period. Demographically, however, little besides race and relationship commitment distinguishes young women who become concurrent from those who do not. Taken together, these findings indicate that sexual concurrency is deeply embedded within young women’s sexual beliefs, norms, and history, and in their relationships with their friends, family, and partners. Interventions aimed at reducing unwanted health outcomes associated with concurrency, such as STI transmission, must either take a broad approach to reducing concurrency, such as using radio or television ads that can reach a large audience,^[57] or focus their efforts on increasing consistent and effective contraceptive use, especially condom use.^{[9, 58]}