Premarital partnerships among youth, including those not involving sexual intercourse, are widely discouraged in India; yet, despite strict sanctions, including parental violence, loss of reputation and swiftly arranged marriages to someone other than the romantic partner, up to 10% of young women and 15–30% of young men form such partnerships.1–6 A review of the literature suggests that little is known about the nature of these relationships,1 such as whether they are romantic or casual, or how individual, family, peer and community factors are associated with partnership formation.

This article aims to address these gaps in evidence by describing the romantic and sexual relationships of unmarried youth in one slum and one rural setting in Pune, India, and by identifying factors associated with their formation. The findings are intended to inform program planners about the range of romantic and sexual relationships experienced by young women and men, and thereby contribute to the design of evidence-based interventions to reduce risky behaviors in this population.


Evidence on factors associated with sexual relations among unmarried youth is sparse. Indeed, a global review concludes that the identification of factors that inhibit or facilitate safe sexual behavior among young women and men is an urgent issue for research.7

The research that is available, which comes largely from developed countries, has identified a number of factors associated with risky or safe premarital sexual relations. Key factors relating to the individual that appear protective against unsafe sex include skills in problem solving, decision making and negotiation, and feelings of self-worth;8–11 on the other hand, substance use and exposure to pornographic materials have been found to be inversely associated with safe sex.10,12

The literature suggests, however, that although individual attributes are important in preventing negative outcomes, a supportive environment—particularly a young person's family, school and peer network—is equally important.8 Youths' behaviors are also known to be influenced by peer-related factors, such as the norms and behaviors of their friends.9,10,12,13 Although peer norms and peer pressure may be important factors underlying early sexual initiation,9,10 it is also likely that larger peer groups and higher levels of social interaction lead to greater opportunities for youth to form romantic partnerships. Finally, among family and household factors, poverty clearly exacerbates the risks that youth face; it constrains their ability to engage in safe sexual relationships and is associated with adverse sexual and reproductive health outcomes.14 Key family factors observed to be protective against unsafe premarital sex include living with both parents, having a father present in the household, appropriate monitoring and supervision, parent-child communication and connection, and the absence of family violence.9,10,13–17

In India, studies that address correlates are rare. Findings from a study of college students in Mumbai, for example, found that individual factors—notably access to resources, attitudes favorable to premarital sex and exposure to pornographic materials—were associated with premarital sex.4 At the family level, individuals who perceived their family environment to be restrictive or uncomfortable were more likely than others to report sexual experience. Findings from a qualitative study of youth in a Delhi slum setting describe how, despite strict parental supervision, girls found ways of forming romantic friendships and engaging in sexual relations.2 Finally, frequency of peer interaction was positively associated with sexual experience.4

Although the existing research has focused on the correlates of premarital sexual behavior among youth, we hypothesize that in a highly restricted social setting such as India's, any partnership—regardless of whether it involves sex—may be influenced by a similar set of factors. Our conceptual framework, which is derived from the available literature, postulates that whether a romantic partnership is formed and whether that partnership involves physical intimacy or sex depends on three sets of factors: individual attributes, such as schooling, economic activity status and agency; familial attributes, such as parent-child communication, parental restrictiveness and family violence; and peer attributes, including frequency and depth of peer interaction and youth norms.

Study Setting

The study was conducted in two sites in Pune District of Maharashtra State. Maharashtra is highly developed and provides youth with a range of educational and employment opportunities.18 Educational attainment levels are considerably higher in Maharashtra than in India overall, and gender differences in enrollment are considerably narrower. In 1998–1999, for example, 9% of 15–19-year-old males could not read or write, compared with 18% of females in that age-group; among 15–17-year-olds, 65% of males attended school, compared with 54% of females.19 Furthermore, marriage occurs during adolescence for large proportions of young women but for few young men: As recently as 2005–2006, 40% of women aged 18–29 were married by age 18; in comparison, just 13% of men aged 21–29 were married by age 21.20

Pune District has a population of 7.2 million and is one of the most economically developed of Maharashtra's 35 districts.18 It is also one of six districts in the state with high HIV prevalence.21 Pune is close to the state's capital, Mumbai; perhaps for that reason, its youth have greater access to education, employment opportunities, modern consumer goods, new ideas and modern lifestyles than youth elsewhere in the state.18,22

Given the sensitivity of our topic, we were concerned about potential community hostility and difficulty in gaining access to youth, especially adolescent females. Thus, we opted to locate the study in settings in which health and development agencies already enjoyed considerable rapport with and insight into the communities they served. In 2001, Mawal Subdistrict had a rural population of 305,083 spread over 180 villages; half of these villages made up the rural setting for our study. The social and demographic profile of rural Mawal was fairly typical of the district more generally: For instance, the scheduled caste and tribal populations comprised 15% each of the district and subdistrict populations, respectively; and female literacy rates were 46% and 52% in rural Mawal and rural Pune, respectively. Pune City—the urban study setting—has a population of 2.5 million people, of whom some 530,000 are estimated to reside in slums.22,23 Our study was conducted in one large slum setting.

The two nongovernmental organizations in whose sites the study was undertaken offered general health and development services, but did not include specific services for sexual and reproductive health of youth.*


Study Design

The study consisted of three phases: a presurvey qualitative phase; a survey of married and unmarried young women and men; and postsurvey in-depth interviews with selected survey respondents.24,25 The presurvey qualitative phase—conducted in 2003—comprised 13 focus group discussions with youth;21 interviews with local health care providers, teachers and other individuals familiar with youth; and 13 in-depth interviews with selected youth. This phase was intended to inform the development of the survey instrument and gain insight into the romantic and sexual experiences of youth. Postsurvey in-depth interviews, 149 in all, were conducted with married and unmarried youth reporting various romantic and sexual experiences in the survey. These interviews were intended to enable researchers to better understand the nature of these relationships. In this article, we restrict our discussion to unmarried participants and focus largely on survey findings, using textual data to provide insight into survey findings.

In light of the expected gender differences in sexual experience, the possible refusal and nonresponse to sensitive questions related to sexual experiences and a probable design effect that could have biased the randomness of the sample, our sampling strategy called for a sample size of 2,150 and 950 unmarried young women and men, respectively, from each site. An initial house-listing exercise identified all households in each site with youth aged 15–24: A total of 21,179 rural households and 19,336 urban slum households, containing populations of 99,237 and 96,680 people, respectively, were enumerated. We prepared lists of all unmarried and married youth. These lists contained 3,646 females and 7,352 males in the rural setting and 4,169 and 8,191, respectively, in the urban setting. Samples were drawn randomly from each list; if a household contained more than one eligible male, only one was selected at random; if more than one eligible female was identified, likewise, only one was selected at random. More than 80% of each of the four sample targets of unmarried females and males in rural and urban slum settings was reached.

Conducted in 2004–2005, the survey used closed-ended questions to explore in detail the romantic partnerships in which young people engage. Questions were posed in terminology identified in the presurvey qualitative phase as language used by young people. Efforts were made to ensure that the interviews were conducted in private and in locations that were convenient for the respondents; interviews typically lasted an hour.

The reliability of survey self-reports on sexual experience is a matter of concern, and we had no way of assessing internal or external consistency of data.26 Thus, we undertook several measures to reduce potential underreporting and made special efforts to build rapport between the study communities and the study team. For instance, emphasis was placed on the selection and training of interviewers: Interviewers, who were generally in their 20s, underwent an intensive three-week training during which they not only became familiar with the questionnaire and with sexual and reproductive matters relating to youth, but also learned how to build rapport with their respondents and overcome any discomfort they had about discussing sexual matters.

The survey instrument was informed by findings from the presurvey qualitative phase as well as other instruments relating to youth behaviors,27–31 and underwent several revisions following pretesting. Questions on the nature of relationships proceeded in a gradual way from less to more sensitive; experiences (e.g., holding hands, kissing and sex) in romantic relationships were asked separately from other experiences—sex that was forced, with a same-sex partner, or, for young men, with a sex worker or married woman. At the conclusion of the interview, respondents were asked a single question: "Have you ever had sex with anyone?" The respondents were asked to indicate their answer on a blank card and place the card in an envelope, which they sealed and returned to the interviewer; they were informed that the principal investigators (but not the interviewers) would be able to link their responses in the face-to-face interview with that reported in the sealed envelope24 Finally, textual data, drawn from presurvey focus group discussions and key informant interviews, enabled researchers to corroborate the evidence presented in survey data.


•Outcome measures.

We included three key outcome indicators. The first measures whether or not young people had ever had a romantic relationship. Those who reported a romantic opposite-sex relationship were asked whether they had ever held hands, hugged, kissed on the lips or had sex with that partner. Respondents were also asked whether they had experienced sex in other situations. Key indicators included whether or not the respondent had ever had a premarital romantic relationship with a person of the opposite sex, whether they had experienced any physical intimacy with a romantic partner and whether they had engaged in sexual relations with a romantic partner or in a number of other situations.

•Individual characteristics.

Key indicators included respondents' age, educational attainment and economic activity, and their exposure to alcohol, drugs and pornographic films. In addition, youth were asked a range of questions concerning their ability to have a voice in determining their own lives, including such dimensions of agency as mobility and sense of self-worth. To measure mobility, we created an index that summed the number of places out of a total of five—a local shop, a friend's house, a film or fair (mela), a temple or mosque, and anywhere outside the neighborhood—young people said they could visit without obtaining permission. Scores on this index ranged from 0 (for those who required permission to visit all five places) to 5 (for those who could visit any of the five places without permission). Self-worth was measured by two dichotomous variables: whether respondents felt that their opinions were respected by their family and whether they found it easy to build new friendships.

•Peer influence and connections.

We included four measures of peer influence and connections: an index of youth norms regarding the acceptability of premarital sex in the urban neighborhood or village of residence, membership in a social group, peer contact and peer support. The youth norm index was based on youth agreement or disagreement with four statements: that it was all right for young men and women to kiss, hug and touch each other; that there was nothing wrong with engaged couples having sex before marriage; and that it was all right for young men and young women, respectively, to have sex before marriage. Respondents were assigned a score of 1 if they agreed with the statement and zero if not.

To construct the youth norm index, we first summed responses of all individuals to the four statements concerning the acceptability of premarital sex. We then calculated the average for each gender in each village or urban block. Each respondent residing in a specific village or urban block was assigned the mean value for males or females in that community; the index ranged from 0 to 4, with 0 signifying that youth in a community considered premarital sex unacceptable in all four situations and 4 signifying that youth considered it acceptable in all four situations.

We measured membership in a social group (mandal) by a dichotomous indicator. Peer contact was measured by an index of the frequency of respondents' interaction with peers, ranging from 0 (never) to 3 (regularly). Finally, peer support was measured by an index of the extent to which respondents identified peers as their most likely confidants on life issues. Youth were asked with whom (among a range of options including peers, parents, siblings, extended kin, teachers and so on) they would be most likely to discuss issues related to health, education, work, relationships with the opposite sex, menstruation (young women) or nocturnal emission (young men), and family matters. A score of 1 was assigned for each of the six types of issues for which peers were identified as the most likely confidants; thus, the index ranged from 0 to 6.

•Family influences.

We included dichotomous measures of residence in a two-parent household, observation or experience of family violence, paternal substance use and perceived parental strictness. In addition, we constructed an index of parental support—identical to the index of peer support discussed above—that summed how many times respondents identified a parent as their most likely confidant on six types of life issues; the index ranged from 0 to 6. In addition, background characteristics of youth were collected, including, for example, religion, housing amenities, household economic status and parental educational attainment.

Data Analysis

We first calculated the proportions of young males and females in each setting (urban or rural) who reported aspects of romantic and sexual partnerships, and the proportions reporting selected individual, peer and family characteristics. In each case, t-tests were calculated to identify significant differences between young women and men in each setting.

Finally, we conducted multivariate logistic regression analyses to identify associations between selected individual, peer and family factors and youths' relationship experience. The first analysis examined romantic partnerships of males and females with a member of the opposite sex. Because few young women reported sexual intercourse, the second analysis focused on young women's experience of physical intimacy with a romantic partner (defined as holding hands, hugging, kissing on the lips or sexual intercourse) or sex with any other partner. In the third analysis, the dependent variable was young men's experience of sexual intercourse with any partner (i.e., romantic partner, sex worker or other). Data were weighted to reflect the rural-urban distribution of Pune District.


Respondent Characteristics

In all four groups, the vast majority (83–92%) of the young people who responded to questions on partnerships were Hindu (Table 1). Economic status, as measured by the mean number of seven consumer goods owned (TV, telephone, pressure cooker, mobile phone, bicycle, motorcycle or car and VCR), appears to be similar across groups (2.7–3.3), although rural households typically possessed slightly fewer amenities than urban households. A large majority of urban and rural youth (87–98%) reported that their households had electricity, but a larger proportion of urban respondents than of rural respondents cooked with gas (80–86% vs. 45–51%) or had access to piped or well water within the home (69–87% vs. 35–38%). In contrast, a larger proportion of rural youth than of urban youth reported having toilet facilities in their home (21–27% vs. 7–8%).

Partnerships and Sexual Relations

About one-third of urban males and females and one-fourth of rural males and females reported having made or received a proposal of romantic partnership (Table 2, page 154). Within each setting, a larger proportion of young men than of young women had made or received such proposals; however, the extent to which proposals were accepted and the nature of subsequent relationships varied widely by the respondent's gender and residence. In the urban setting, 24% of males and 8% of females reported ever having had a romantic partner; the proportions in the rural setting were 17% and 5%. According to findings from a life table analysis using Kaplan-Meier estimates that examined the probability of having spent time alone with the partner by age 16 (not shown), 11% percent of urban males and 7% of rural males had done so; the proportions among females were 6% and 4%.

Overall, 18% of urban males and 16% of rural males reported having had sexual intercourse, compared with 1% and 2% of females. When we probed about type of partner, 8–9% of young men and fewer than 1% of females reported having had sex with a romantic partner. Some 9% and 3% of urban and rural young men, respectively, had had other sexual experiences, including sex with another man, exchange or paid sex, forced sex or sex with an older married woman; a very small proportion of young women (fewer than 1%) reported forced sex, exchange sex or same-sex relations. Also, a small proportion of young men—particularly in urban areas—who reported sex in the face-to-face interview reported sex with both a girlfriend and at least one other partner. In addition, 3–5% of young men and 1–2% of young women reported by sealed envelope (but not in their interview), that they had experienced sexual relations. Some 20–26% of young men and 4–6% of young women reported ever having experienced any kind of physical intimacy with a romantic opposite-sex partner or sexual intercourse with anyone.

When we considered only respondents who had had a romantic partnership, 85% of young men and about 60% of young women reported having experienced any type of physical intimacy, including holding hands, hugging, kissing or sex; smaller proportions had ever kissed or had sex (76–77% of males and 36–40% of females) and still smaller proportions had had sex (37–49% of males and 6–13% of females). Young men were typically 1–3 years older than their partners.

A smaller proportion of young men than of young women had entered their current relationship with expectations of marriage. This finding was supported by data from in-depth interviews:

"No, not at all [did I want to marry her]. For time-pass, I got friendly with her."—19-year-old urban male

"I asked him if he would get married to me or not. He would only say yes.…When I would ask him anything, he would say that when the time comes, we shall talk.…I would think that he did not want to get married to me because whenever I would talk about it he would change the topic and if he wanted to get married to me, he himself would have brought up the topic."—17-year-old rural female

Finally, although 82–85% of youth in a partnership reported that their peers were aware of the relationship, only 30–49% said that their parents knew about it.

Individual, Peer and Family Factors

•Individual characteristics.

On average, young men in the sample were older than young women (19 years vs. 17 years—Table 3). A large majority (73–81%) of young people had completed eight or more years of school. In the 12 months preceding the survey, 64–67% of males had engaged in a wage-earning activity, compared with 27–34% of females.

There was a large gender difference in mobility, with young males reporting greater freedom of movement than young women (mobility index score, 3.0–3.2 vs. 1.6–1.7). Gender differences in indicators of self-worth were mixed. For instance, 92–95% of males and 88–91% of females reported that their views were respected by their family; however, confidence in one's ability to forge new friendships was reported by greater proportions of females (30–31%) than of males (18–25%).

Finally, large gender differences were apparent in exposure to alcohol, drugs and pornographic films. Fifty-seven percent of urban males and 41% of rural males reported having used alcohol or drugs, or having been exposed to a pornographic film, compared with 2% and 1% of females, respectively.

•Peer influence and connections.

Mean youth norm index scores were low among young men (0.6–0.7—Table 4) and lower still among young women (0.1–0.2), indicating that norms about premarital sex remain traditional overall. Large gender differences in aspects of young people's peer networks were evident. A much smaller proportion of young women than of young men reported being a member of a social group (4–5% vs. 44–56%), and young women reported having fewer friends than did young men (averages of five and 15–19 friends, respectively). Furthermore, young women scored lower, on average, than young men on the peer contact index (0.7–0.8 vs. 1.7–1.8) and peer support index (0.7–0.8 vs. 1.6–1.7), suggesting that young women are in touch with and confide in their friends less frequently.

•Family and parental influences.

At least 74% of youth resided in two-parent homes, and more than 90% lived with at least one parent (Table 5, page 156). On average, youth were better educated than their parents: Whereas youth had completed 9–10 years of school, their fathers had had 5–6 years and their mothers 2–3 years.

Clear gender differences were observed in youth perceptions of parental strictness, underlining the greater permissiveness with which young males are socialized. Although 40–41% of young women reported their parents to be strict, only 28–32% of males did so. This finding is reinforced by data from focus group discussions. For example, an urban female said, "If we start a friendship with a boy, if he meets us somewhere and if our parents see us talking then something might come to their minds… parents are really very strict."

A sizeable minority of respondents depicted their family life as involving violence and substance use. Overall, 12–19% of youth reported having witnessed their father beating their mother, with little difference by gender; 8–15% of young women and 22–32% of young men reported having been beaten by a family member, usually a parent. Some 28–45% of youth reported paternal substance use (mainly alcohol).

On average, young women scored higher than men on the index of family support (4.3–4.4 vs. 2.5–2.7); however, both genders confided about more matters to family than to peers. The exception was problems in relationships with the opposite sex: Some 45–63% of youth confided in peers about such matters, whereas 16–27% confided in parents. In addition, young men tended to confide more in peers about nocturnal emission, whereas young women confided more in their mother about menstruation.

Partnership Progression: Correlates

In multivariate analyses, factors from all three levels—individual, peer and parental—were found to be associated with romantic partnerships and physical intimacy for females; only individual and peer factors were linked to sexual relations for males (Table 6, page 157). Several individual factors were associated with partnership formation for both young women and young men. For example, age was positively associated with the odds of having had a romantic partnership for females (odds ratio, 1.1) and of having had sex for males (1.1); education was negatively associated with the odds of forming a romantic partnership or of engaging in physical intimacy for females (odds ratio, 0.9 each), and of having sex for males (0.9). Exposure to alcohol, drugs or pornographic films was positively associated with all outcomes, for both young men and young women (3.5–5.7).

Differences by gender, however, were also apparent: Among young men but not young women, current wage work was associated with having formed a romantic partnership (odds ratio, 1.5) and mobility was associated with having experienced a sexual relationship (1.1). In contrast, women who perceived themselves as being able to make friends easily had elevated odds of having had romantic and physical relationships (1.6 for each); among men, this factor had only a marginally significant relationship with the formation of romantic partnerships (1.4).

Of the peer-related factors, greater community-level acceptance of premarital sex was strongly associated with romantic partnership formation for males and females, physical intimacy for females and sex for males (odds ratios, 2.1–5.2), as was greater frequency of peer contact (1.3–1.6). The influence of other peer factors was less consistent. Group membership was strongly associated with romantic partnership formation for males (odds ratio 1.6); peer support was associated with physical intimacy for females (odds ratio 1.2).

Qualitative data confirmed that in the area of romantic and sexual relations, peers—particularly for young men—offered support in many ways, ranging from providing space where a couple could meet to giving advice on appropriate methods of contraception, as revealed in the following:

"We have friends with whom we discuss everything…. And having such friendships is necessary since we discuss such things [romance, sex] only amongst friends. We never discuss such things with family members."—focus group discussion, urban males

"[We had sex] there only—in my friend's house. She [the friend] used to wait outside or would go somewhere." —in-depth interview, urban female, age 16

"My friends told me that if a condom is used then the girl does not get pregnant."—in-depth interview, rural male, age 21

In contrast, family influences and connections were more likely to be associated with outcome measures for young women than for young men. For example, young women who reported having a close relationship with their parents had reduced odds of forming romantic relationships (odds ratio, 0.9), while those who had witnessed their father beating their mother had elevated odds of forming such relationships (1.6). Young women who had been beaten by their family were significantly more likely to form romantic (3.0) and sexual (2.6) partnerships.

In focus group discussions and in-depth interviews, young women and some young men themselves made the link between parental support and partnership formation. For example:

"Why does a boy or a girl go out? It is when there is no one at home who understands him or her.…Then with whom should they share their problems? When he faces this problem, then he starts searching for a solution [opposite sex friend] in the neighborhood."—focus group discussion, urban males

"Sometimes if in the family, someone is not getting love, or if every day the father comes home drunk…in such situations one tends to find some support outside."—in-depth interview, urban female, age 23

In regard to parental monitoring, youth who reported parental strictness were no less likely than others to report a relationship that included any physical intimacy (females) or to have engaged in sex (males). Textual data suggest that opportunities do arise, despite strict supervision, for youth to find moments of privacy:

"I told her that whatever they do after marriage, I wanted to do it only once. In the beginning she did not agree but afterwards she told me that on Saturday afternoon her family was going out. She told me that she would be at home."—in-depth interview, rural male, age 20

"When I used to go to school he used to come behind me. Sometimes he used to meet me and tell me [where to meet]. So I used to tell my parents I had class or some programme in school."—in-depth interview, 19-year-old urban female


Our study builds on previous research and makes several new and important contributions to understanding the levels and correlates of premarital relationships among youth. Our findings are community-based and reiterate that even in this traditional setting, opportunities do exist for social mixing between young men and women, and that expressions of interest in developing intimacy with the opposite sex are not uncommon. Opportunities also exist for the formation of romantic partnerships among unmarried youth, for physical intimacy (including sex) within such partnerships and for sex with romantic and other partners. Overall, 16–18% of young men and 1–2% of young women reported having had sex—levels that fall within the broad range reported in other smaller and less representative studies conducted in India.1–6

In addition, our study adds to the limited body of research on the associations between individual, peer and family factors and the formation of premarital romantic partnerships and engagement in premarital physical intimacy or sex.4,8–17 Our findings, like those of previous research,4 suggest that age, education, frequency of peer contact, positive attitudes toward premarital sex and exposure to alcohol, drugs or pornographic films are indeed correlated with romantic or sexual experience for both young women and young men.

Our data also show gender differences in correlates of premarital relationships. For young women, romantic relationships were negatively associated with closeness to parents and positively associated with having seen their father beat their mother. Young women who had been beaten by their families had elevated risks of forming romantic and physically intimate relationships. These correlations did not appear among young men.

Although peer influences were associated with increased odds of physical intimacy and sex among young women and men, the associations tended to be stronger among young women. Peer support may enable closely guarded young women to defy traditional norms, to exercise individual choice and to engage in nonconformist behavior, whereas young men face fewer restrictions and therefore are less likely to require peer support to engage in nonconformist—including sexual—behavior. Also, our data suggest that young men's greater mobility and greater access to resources enable them to form romantic or sexual relationships; in contrast, among young women, it is clearly those who express agency—in terms of the ability to forge new friendships—who are able to overcome parental restrictions on mobility and form relationships.

Finally, our finding of a marginal, positive relationship between the strictness of socialization and engagement in romantic partnerships and the lack of association with physical intimacy or sex suggests that parental beliefs that strict supervision of children may inhibit their formation of romantic or sexual relationships may be unfounded. On the other hand, as seen previously in the literature,2 closeness to parents may discourage the early formation of romantic relationships for young women, while family violence may raise the risk of all types of relationships.


Potential limitations of our study must be acknowledged. First, youth in our sample may have underreported their romantic, physical and sexual experiences—a limitation observed in most studies of this nature.7,26 Pune is a traditional setting, where powerful norms inhibit premarital friendships with the opposite sex, whether they are platonic, romantic or sexual. In such a setting, relationships are usually carried on secretly, and youth—particularly young females, who have more to lose than young males—may be unwilling to disclose them. And although young males may have been more forthcoming than young females in their reporting of romantic partnerships, they may have underreported their same-sex or transactional sex experiences. We attempted to reduce underreporting by arranging the progression of survey questions from less to more sensitive, and by giving respondents the opportunity to report sexual behavior anonymously. Reporting in our study is consistent with that in other studies in India,1-6 few of which extended their samples to youth at the lower end of the age spectrum (i.e., 15–16). Thus, although we cannot rule out the underreporting of premarital experiences and possible measurement error, we believe that they are unlikely to be of sufficient magnitude to compromise the validity of our findings.

A second potential limitation is the cross-sectional nature of our study and the resulting inability to infer causation. Our study findings cannot be interpreted as evidence of the determinants of romantic partnerships, physical intimacy or sexual relations; however, in some cases—notably parental factors—temporal ordering rules out the possibility that causality could go in either direction. Qualitative evidence strengthens the suggestion of a causal link between some of the associations observed in multivariate analyses—for example, between family instability or closeness, the supportive role of parents and engagement in romantic or physical relationships.

Program Implications

The declining age at puberty and the increasing age at marriage have created a growing period in which young people may engage in premarital romantic and sexual relations. Likewise, evidence that large proportions of youth remain in school for extended periods suggests that opportunities to spend time together in acceptable spaces away from the watchful eyes of parents will increasingly present themselves. The challenge is, therefore, for programs to ensure that young women and men are fully informed and equipped to make safe choices and negotiate wanted outcomes. Sexuality education must be made universal, and should address relationship issues as well as consent and safety from an early age in schools and other settings in which young people congregate.

Our findings also highlight the importance of peer networks, which can have a positive or negative influence on the safety and wantedness of young people's relationships. At present, it appears that the network plays a role in enabling the establishment of romantic or sexual relations. Good sexuality education may enable peers to play a role in ensuring that these are safe and wanted as well.

Also needed are efforts to ensure a supportive environment; programs need to address parental inhibitions about discussing sexual matters with their children and encourage greater openness and interaction between parents and children. Finally, gender disparities in the extent and correlates of premarital partnership formation and experience of sexual relations argue against generic programming. They suggest a need for sexuality education programs tailored to the different circumstances and experiences of young women and men; and for programs that sensitize parents about more equitable socialization patterns and ways of developing closer interaction with both daughters and sons.