Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Sep 29.
Published before final editing as: J Marriage Fam. 2026 Aug 4:10.1111/jomf.70111. doi: 10.1111/jomf.70111

Correlates and Patterns of Adolescents’ First Romantic Relationships in Three Low- and Middle-Income Countries

Melissa Alcaraz 1, Sarah R Hayford 2, Natalie D Eggum
PMCID: PMC13618987  NIHMSID: NIHMS2204071  PMID: 42807724

Abstract

Objective:

This study characterizes adolescent romantic relationships in three low- and middle-income countries (LMICs) with varied and evolving family systems.

Background:

Adolescent relationships matter for young people’s future outcomes and development. These relationships vary widely, from short-term, casual relationships to more sustained and emotionally engaged relationships. They can also serve as precursors to family formation as young people approach adulthood, particularly in contexts where companionate marriage is normative and where the age at marriage and childbearing has increased. However, there is limited empirical research focused on adolescent romantic relationships outside of high-income countries in North America and Western Europe.

Method:

Using data from the FAMELO project, we examine the covariates and characteristics of adolescents’ first romantic relationships in Jalisco, Mexico; Gaza province, Mozambique; and Chitwan Valley, Nepal. We use logistic regressions to examine the likelihood of adolescents reporting relationship experience and latent class analysis (LCA) to characterize adolescent romantic relationship types.

Results:

Compared to peers without relationship experience, adolescents who engage in romantic relationships are more likely to be older, male, feel more like adults, and report lower caregiver supervision. LCA results show substantial variation in the prevalence and types of romantic relationships that adolescents experience in distinct international contexts.

Conclusion:

Our findings highlight the importance of studying variation in adolescent romantic relationships, particularly in LMIC contexts, for more fully understanding shifting family formation trends.

Research in the United States and other high-income industrialized countries consistently documents that adolescent romantic relationships are both common and consequential (W. A. Collins et al., 2009; Giordano, 2003; Meier & Allen, 2009; Olmstead, 2020). They help shape adolescents’ mental health and socioemotional development (Kansky & Allen, 2018; Villalta et al., 2023). They also serve as precursors to potentially more serious relationships as youth approach adulthood (Meier & Allen, 2009), particularly in contexts where companionate marriage is normative and where age at marriage and childbearing has increased (Allendorf, 2017; Manning et al., 2008). However, adolescent relationships remain relatively understudied and undertheorized in low- and middle-income countries. Existing research on low- and middle-income countries is disproportionately focused on risky sexual behavior as well as early and/or unintended childbearing (but see, e.g., Ahearn, 2001; Bingenheimer et al., 2017; Grant et al., 2013; Poulin, 2007, for key exceptions).

In this Brief Report, we begin to address this gap by using unique data to analyze adolescents’ first romantic relationships in sites in Mexico, Mozambique, and Nepal. We use survey data from the Family Migration and Early Life Outcomes (FAMELO) project, a multi-site, household-based survey, to characterize first relationships reported by adolescents ages 14 to 17 in three low- and middle-income countries (LMICs) with varying cultural and family systems. We use logistic regressions to examine the individual- and family-level factors associated with adolescents reporting relationship experience. We also use latent class analysis to identify profiles of adolescent relationships. We find distinct correlates and types of adolescent relationships across contexts, highlighting the need for contextually situated research on these early stages of the family formation sequence.

Adolescent Romantic Relationships in Developmental and Cultural Context

Adolescence is a critical developmental stage spanning the period between childhood and adulthood. In addition to the onset of puberty, this time is characterized by identity development, increased independence from the natal family, increased involvement in decision-making, and transitions in roles and responsibilities (Elder, 1998; Johnson et al., 2011). In many contexts, adolescence is also the period when people enter romantic relationships for the first time. The nature of adolescence is structured by educational opportunities, economic chances, and social and cultural expectations (Clark & Agnant, 2025; National Research Council and Institute of Medicine, 2004; Settersten Jr. et al., 2008). For example, in contexts where education is more widely accessible and available, young people may have higher levels of educational persistence and progression, which may lead them to be more economically stable and delay family formation (Clark & Agnant, 2025). Adolescence is also shaped by the social and cultural expectations for appropriate behavior within respective cultural contexts (e.g., dating practices, working for pay, etc.) (National Research Council and Institute of Medicine, 2004). The developmental feature of adolescence and the transitions considered as important in the life course, such as school leaving, residential independence, individual autonomy, and family formation, draw their meaning from particular economic and social systems (Gibbons & Poelker, 2019; Juárez & Gayet, 2014).

In contexts where marriage takes place earlier in the life course, adolescent relationships may be seen as a direct preparation for or precursor to marriage, while in contexts with later marriage, adolescent relationships may be understood primarily as fun or recreational. When marriage occurs later, there is more time in adolescence and early adulthood for different kinds of relationship experiences – and these experiences may become more normative (W. A. Collins et al., 2009; Manning et al., 2008). Where parents and families are heavily involved in choosing marriage partners, adolescent relationships that reflect individual decision making may be viewed as unnecessary or even harmful. In contexts where age at first marriage is rising, romantic relationships in adolescence will likely become increasingly common and carry more significance as young people transition to adulthood (Clark & Mathur, 2012; Connolly et al., 2023; Manning et al., 2008). This is particularly relevant in our study contexts of Mozambique and Nepal, where young people are gaining more power in the marriage decision-making process, as evidenced through the growth of choice marriage and through the reduced expectation of bridewealth payments (Axxe et al., 2024; Chae et al., 2021). Given these changes, adolescent dating patterns may provide a window into family formation processes in contexts undergoing significant and rapid social change.

Variation in the Nature of Adolescent Romantic Relationships

Collins (2003) proposed five key domains for characterizing adolescent romantic relationships: involvement, partner selection, relationship content, quality, and cognitive and emotional processes. “Involvement” indexes whether adolescents are engaged in relationships, while “partner selection” refers to both the process (how partners meet) and criteria for selection (who is desirable). “Content” is the activities that partners engage in together; “quality” includes intimacy, affection, and nurturance; and “cognitive and emotional processes” incorporate aspects such as expectations for how the relationship will proceed and goals for the relationship.

Because expectations and experiences of adolescent peer relationships, relationships with family, and socioemotional development are dependent on social and cultural context, adolescent romantic relationships would also be expected to vary within different national contexts (Cavanagh, 2007; W. A. Collins et al., 2009). In the United States, adolescent relationship characteristics and dynamics vary for different race-ethnic groups, which has been interpreted as reflecting the importance of cultural context to relationship behavior (Giordano et al., 2005; Gowen et al., 2014; Laborde et al., 2014; Meier & Allen, 2009). However, little empirical research directly documents variation in adolescent romantic relationships in contexts outside the United States and Western Europe. In low- and middle-income country contexts, there has been a large focus on family planning, reducing child marriage, and increasing contraceptive access over the last half century (Ankomah, 1999; Campero et al., 2011; Clark & Mathur, 2012; Grant, 2012). As a result, research has focused on the association between adolescent sexual behavior, pregnancy, and other outcomes (Clark et al., 2010; Luke, Goldberg, et al., 2011; Vargas Trujillo & Barrera, 2002). The lack of research on adolescent relationships in LMICs is a notable omission, given that to fully understand existing and shifting trends in family formation, a fuller understanding of modern courtship and dating practices (e.g., adolescent relationships) is needed.

Study Settings

The three study settings represent different social and family systems, cultural contexts, and levels of economic development (e.g., participation in the formal economy, infrastructure development, etc.). They differ in the acceptability and prevalence of adolescent relationships, as well as the timing and ordering of family formation (i.e., dating, childbearing, marriage, etc.). Across all three contexts, the mean age at first marriage has increased over time for both men and women, but women enter their first unions at younger ages than men (Batyra et al., 2021; Bongaarts et al., 2017; Marphatia et al., 2020; Ortíz-Ávila, 2020). We briefly summarize additional key aspects of the family demographic context for each study site. Collectively, existing evidence on family formation and demographic events suggests that adolescent romantic relationships may play an increasingly important role in understanding family systems.

Mexico.

Expressing romantic affection through dating, exclusive relationships, and physical affection is a relatively normative experience for Mexican adolescents (Espinosa-Hernández & Vasilenko, 2015). In Mexico, marriage and childbearing are increasingly decoupled (Ortíz-Ávila, 2020). In 2023, the average age at marriage for people in heterosexual unions was 32 for women and 35 for men (INEGI, 2024b). These averages mark notable increases over the last decade, as the average age at marriage for women was about 28 years old in 2014; in the same year, the average age at marriage for men was about 30 years old (INEGI, 2024a).

Mozambique.

Romantic involvement in adolescence is common and accepted for Mozambican adolescents (Archambault, 2017; Parkes et al., 2016). Bridewealth marriage is becoming less common in Mozambique, even as early marriage remains a common response to teenage pregnancy (Chae et al., 2021; Kok et al., 2022). In 2019, the mean age at first marriage for women was approximately 19 years old (World Bank, 2025). According to the 2022–23 Demographic and Health Survey in Mozambique, the median age at first marriage in Gaza Province was 19.1 years for women and 23.1 years for men, approximately the same as measured in the previous DHS survey in 2011 (Instituto Nacional de Estatística (INE) e ICF, 2024; Ministerio da Saude (MISAU) et al., 2013).

Nepal.

Adolescent dating relationships have been highly stigmatized in Nepal until relatively recently (Ahearn, 2001); consequently, adolescent relationships are uncommon and little is known about these relationship experiences. Youth in Nepal experience the most strict control over issues of sexuality across our three sites (Allendorf, 2017; Axxe et al., 2022), and families play a central role in choosing spouses. Women’s perceived sexual purity remains an important prerequisite for favorable marriages, and romantic relationships outside of courtship or marriage remain highly stigmatized (Allendorf, 2017; Desai & Andrist, 2010). Women’s median age at first marriage is about 18 years old, and men’s median age at first marriage is about 22 years old, reflecting recent increases. Median age at first marriage in 1996 was about 16 years old for women; in 2001, median age at first marriage for men was about 20 years old (Ministry of Health and Population [Nepal] et al., 2023).

Current Study

In this Brief Report, we bring together developmental and sociological perspectives to examine romantic relationships as part of the family formation sequence in the early life course, focusing on the structural and contextual factors associated with relationship experiences. We are guided by two general research questions. Our main research question is: how do (a) the distribution and (b) the nature of adolescent romantic relationships vary within different contexts based on individual and family characteristics? We conduct country-specific analyses to address this research question. Secondarily, we consider how the nature of adolescent romantic relationships varies across contexts based on the nature of social and family systems. In the conclusion, we compare results across countries to address this latter research question.

We first examine the likelihood that adolescents report relationships to understand how individual and family characteristics are associated with relationship experience. Individual-level correlates of relationship involvement include age and gender, as primary drivers of developmental patterns (Furman et al., 2019; Lantagne & Furman, 2017); adolescents’ attitudes and future goals, which shape their life course pathways and thus the centrality of romantic relationships in adolescence (Furman & Shaffer, 2003; Meinster & Rose, 2001; Shulman & Connolly, 2013); family resources, which both directly influence adolescents’ opportunities and indirectly shape choices about family formation (Schneider & Hastings, 2015; Sheppard et al., 2016); and family structure and functioning, which guide adolescent choices and determine adolescent openness to romantic relationships (Cavanagh et al., 2008; Halpern-Meekin & Turney, 2022).

We then use latent class analysis to identify distinct types of romantic relationships. Previous research has highlighted considerable heterogeneity in adolescent relationships (W. A. Collins et al., 2009; Connolly et al., 2023; Espinosa-Hernández & Vasilenko, 2015), and latent class analysis allows us to observe potential differences in the content of these relationships holistically rather than studying specific relationship characteristics in isolation (Heuveline & Timberlake, 2004). Because the existing evidence base describing and theorizing adolescent relationships in LMICs is limited, we do not propose formal hypotheses, but we outline some general expectations based on current understandings of changing family systems.

In considering within-country variation, we expect to see different types of relationships in each setting, as characterized by content, quality, cognitive processes, and embeddedness with other social relationships. In considering cross-country context variation, we expect that adolescent relationships will be more common and more varied in settings (e.g., Mexico and Mozambique) where marriage takes place later and where young people have a larger role in choosing their own marriage partners (Kok et al., 2022; Manning et al., 2008), as these factors may be associated with a longer period of dating that extends into adolescence (Allendorf, 2017; Axxe et al., 2024; Chae et al., 2021).

Method

Data

Data come from the Family Migration and Early Life Outcomes (FAMELO) project, a three-country survey data collection project studying health and socioemotional outcomes for children in historically high-migration contexts. Data collection took place between September 2017 and May 2018 in three sites: Jalisco, Mexico; Gaza, Mozambique; Chitwan Valley, Nepal. Sites were chosen systematically to capture contexts with high levels of migration under different migration systems and with differing levels of socioeconomic development and differing family systems. They reflect illustrative case studies rather than a systematic sample of country contexts. All data collection and management procedures were approved by local Institutional Review Boards (IRBs) in each data collection site. Approval was also obtained by the U.S. university IRBs of FAMELO principal investigators.

FAMELO is a two-stage household-based survey. First, communities (villages, towns, or areas) in each site were selected for inclusion. Second, households were sampled from within selected communities. Households were eligible if at least one child ages 5–17 and their primary caregiver lived in the household. In eligible households, one or two children were randomly selected. The number of children interviewed per household was randomly selected. The selected child(ren) and the primary caregiver were both interviewed. Children were asked about future goals, family relationships, schooling, social support, and other topics related to child wellbeing, including peer and romantic relationships. Primary caregivers answered questions about themselves, other members of the household, and the child being interviewed. In Mexico, 2,268 households were sampled, including 3,249 children and 2,268 adult caregivers. In Mozambique, 2,206 households were sampled, including 3,213 children and 2,206 adult caregivers. In Nepal, 2,333 households were sampled, including 3,202 children and 2,333 adult caregivers. Participation compensation procedures were determined in consultation with local interviewers and research personnel to be consistent with local cultural norms. Participating families in Mexico received boxes of nonperishable foods and household products as an incentive for participation. Participants were not compensated in Mozambique or Nepal.

Sample

Adolescents ages 14–17 were asked whether they had had a relationship with a boy/girl who was “more than a friend.” (Adolescents were only asked about different-sex romantic relationships; children younger than 14 were not asked about romantic relationships) There were 873 adolescents age 14 to 17 in Mexico, interviewed in 794 unique households; 712 adolescents ages 14 to 17 in Mozambique, in 681 unique households; and 1,201 adolescents ages 14 to 17 in Nepal, in 1,100 unique households. Of these, 472, 265, and 142 adolescents reported at least one relationship in Mexico, Mozambique, and Nepal, respectively. See Table A1 in the Online Supplemental Appendix for additional information on sample characteristics. In all three contexts, adolescents who responded that they had had at least one romantic relationship were asked detailed questions about their first romantic relationship and most recent romantic relationship (if they had more than one). To capture relationships at a similar developmental stage, we focus on first relationships.

Measures

Adolescent characteristics.

Adolescent demographic characteristics are included as core correlates of potential relationship experience. We control for age and gender. (Gender was self-reported as either male or female; the surveys did not include categories for other gender identities.) In addition to chronological age, adolescents reported subjective feelings of adulthood, measured using a 5-point scale in response to the question “how often do you feel like an adult?”, where 1 = never and 5 = always.

We also included several measures that captured adolescents’ aspirations and plans for the future, because goals related to both family and non-family outcomes are likely to be related to orientations toward romantic relationships. Respondents were asked about the highest level of education that they would like to obtain, which we coded as a three-category measure of low, medium, and high educational aspirations. In Mexico and Nepal, these categories corresponded with the following educational aspirations: less than a post-secondary degree (low), post-secondary degree (medium), and post-graduate education or higher (high). In Mozambique, these categories were lower secondary or less (low), any upper secondary (medium), and any post-secondary education (high). Lastly, we included four variables that capture ideas about gender roles and family formation aspirations. First, adolescents were asked at what age they would like to get married; this variable is measured continuously in years. Second, adolescents were asked how important it was for them to get married one day, and response categories ranged from 1 = not at all important to 5 = very important. Third, adolescents were asked about their desired family size via the question, “If it were up to you to decide, how many children would you like to have in your life?” Desired family size is measured continuously in years. Fourth, adolescents were asked the extent to which they agreed on a scale from 1 = strongly disagree to 5 = strongly agree with the statement, “It is best if the husband makes decisions for the household and the wife follows his decision.” Here, a higher number represented more gender traditional views.

Household and family characteristics.

To ascertain whether experiences in the family of origin are related to adolescent engagement in romantic relationships, we included measures of caregiver education and household wealth. Context-specific measures of primary caregiver educational attainment were recoded into three categories: low, medium, and high educational attainment. In Mexico and Nepal, educational attainment was measured as primary school or less (low), any lower secondary (medium), and upper secondary or higher (high). In Mozambique, the corresponding educational attainment categories were preschool or lower (low), any lower secondary (medium), and upper secondary or higher (high). In Nepal, we used a categorical measure of religious caste, which captures both religious beliefs and socioeconomic circumstances. Caste is measured using three distinct categories: Brahmin Chetri, Hill Janajati, and Other (Newar, Terai Janajati, Dalit, and Other). Lastly, we created country-specific measures of household wealth based on the results from principal components analysis (PCA) within each country, consistent with a widely-used approach in research on LMICs (Corsi et al., 2012; Filmer & Pritchett, 2001) that has been applied in previous work using FAMELO data (Alcaraz et al., 2022; Axxe et al., 2022). The scores are mean-centered based on the country-specific means.

Caregiver supervision and support may shape whether adolescents form romantic relationships and which types (Auslander et al., 2009; Xia et al., 2018; Zimmer-Gembeck et al., 2017). Thus, we account for primary caregiver supervision, mother support, and father support with three individual summative scales created from multiple questions meant to capture each construct; see Table A2 in the Online Supplemental Appendix for a list of all items used to create scales. For example, adolescents were asked to report on a scale from 1 = never to 5 = always, “how often does your [caregiver] usually ask you about your plans for the coming day?”. For each individual scale, responses to all items were summed. See Table A3 in the Online Supplemental Appendix for descriptive statistics (mean, standard deviation, range) and Cronbach’s alpha values for each scale, by country.

Relationship characteristics.

We included measures of multiple aspects of romantic relationships, based loosely on the five domains of relationship characteristics outlined in Collins (2003). These are measured dichotomously, as all except three of the indicators we used (i.e., where adolescents met partners, reasons for starting the relationship, and relationship length) were originally measured via yes/no questions. For consistency across contexts and in prioritizing parsimony, we recoded the three indicators into dichotomous format. See Table A4 for a full description of all relationship characteristics. Involvement is captured by the dichotomous measure of whether adolescents report ever having a relationship, where 0 = “no” and 1 = “yes”. Partner Selection is captured by questions about how adolescents met their partner. Possible responses included at school, in the neighborhood, introduced by friends/family, at a public place, and internet/telephone; we dichotomized this variable as 0 = met somewhere besides school vs. 1 = met at school because meeting at school was the most common response in both Mexico (61.3%) and Nepal (62.0%) and the second most common form in Mozambique (36.2%). (Meeting in the neighborhood (43.0%) was most common in Mozambique.) Relationship Content is operationalized as the degree to which the romantic relationship is integrated with other relationships and how partners spend time together. Adolescents were asked who knew about their relationship (friends, parents, other family), whether they had met their significant other’s family, and the activities they engaged in with their partner (going to parties, spending time together in the community, spending time together outside of the community). Adolescents were also asked about physical intimacy (hugging, kissing, sex).

We do not have direct measures of Relationship Quality; following Collins (2003), we include relationship duration as a proxy for quality. Adolescents were asked about the length of their relationship, in months; for adolescents who responded that their relationship was ongoing, we calculated the length of the relationship at the time of the survey using time-stamped data for the date of the interview and the respondent’s answer to the time when the relationship began. We dichotomized relationship duration as 0 = less than three months vs. 1 = three months or longer. For Mexico, about half of relationships lasted less than three months. Though relationships in Mozambique and Nepal had longer average durations (13.4 months and 16.6 months, respectively), we used three months as a benchmark for consistency across contexts. We further note that relationship duration does not meaningfully distinguish relationships in Mozambique and Nepal, though this distinction does matter for sorting relationships in Mexico.

The last domain, Cognitive and Emotional Processes, reflects adolescents’ goals for and conceptualizations of the relationship. We have two indicators for this domain. First, adolescents were asked what they were looking for when they started the relationship. Possible responses were to have fun, love/companionship, sex, to try out a relationship before marriage, to prepare for having a family, or for financial support. Respondents could choose multiple responses; based on the distribution of responses, we created a dichotomous variable indicating whether the adolescent mentioned love/companionship. The second indicator measures whether adolescents considered the possibility of marriage with their partner at any point in the relationship where 0 = “no” and 1 = “yes”.

Analytic Strategy

We used logistic regression to estimate the likelihood that adolescents ages 14 to 17 reported having ever been in a romantic relationship, separately by country. (See Table A5 in the Online Supplemental Appendix for descriptive statistics for all regression variables.) Adolescents are the unit of analysis, and robust standard errors are clustered at the household-level to account for the fact that more than one child could be sampled from the same household. To handle missing data, we used multiple imputation by chained equations in Stata 18, estimated separately by country context. This multivariate data imputation method produces and combines results from 25 imputed datasets with 100 burn-in iterations for each dataset (Aloisio et al., 2014; Royston & White, 2011); see Table A5 for additional information on missing data.

We then leveraged latent class analysis to examine variation and similarity in relationship characteristics. Here, relationships, rather than adolescents, are the unit of analysis. Latent class analysis (LCA) assumes probabilistic relationships between latent constructs and actual measures. This method assumes that within any given population, there are certain “classes” of observations and that the observed distribution of variables is attributable to distribution across unobserved categories. Here, LCA identifies “types” of relationships with distinct combinations of characteristics. LCA produces multiple forms of output including number of classes and composition of classes (Gaalen & Dykstra, 2006; Hagenaars & Halman, 1989). After estimating models with different numbers of classes, the researcher typically chooses the final model and number of classes based on a combination of statistical and substantive fit (L. M. Collins & Lanza, 2010). We used this method to estimate models that produced classes with different combinations of relationship profiles and the factors associated with membership in each class. Because more than one adolescent could be interviewed per household (and thus, more than one first relationship could be reported per household), we adjusted standard errors by nesting individuals within households in our LCA estimates.

While relationship indicators were largely measured and operationalized similarly across sites, we allowed for variation to better capture country-specific contexts. We used 13 indicators in each country, 11 of which were shared across countries. Across all countries, we used the following binary indicators: met partner at school, parents knew about relationship, family knew about relationship, friends knew about relationship, met significant other’s family, went on trip outside their home community with partner, hugged partner, kissed partner, the relationship lasted at least three months, possibility of marriage, and whether adolescents started relationships for love. In Mexico, we also included indicators of whether the adolescent reporting the relationship ever attended a party with their partner and whether the adolescent spent time in the public square (plaza) with their partner. In Mozambique and Nepal, we included two additional indicators: whether adolescents had ever spent the night together with their romantic partners and whether they had had sex with their partners. These latter two questions were not asked in the Mexico site. Per Wurpts and Geiser (2014), we kept as many indicators as possible in our analyses, while also favoring higher quality indicators. We recoded and cleaned data in Stata 18.

We estimated country-specific LCA models for first relationships. Number of classes for each country model were determined using Akaike Information Criteria (AIC), Bayesian Information Criteria (BIC), sample-size adjusted BIC, Vuong-Lo-Mendell-Rubin Likelihood Ratio Test (LRT), Lo-Mendell-Rubin Adjusted LRT, entropy, latent class probabilities, class sizes, and class interpretability (see Online Supplemental Appendix, Table A6). Our selection of the preferred model for each country was based on judging across the combination of indicators as well as substantive interpretability (L. M. Collins & Lanza, 2010). We do not formally test for between-country variation, as we prioritize understanding romantic relationships within each context’s unique gender and family systems (our primary research question). LCA models were estimated in Mplus 8.

Alternate specifications.

For estimates from alternative LCA models (i.e., only shared indicators, stratified by order of relationship), see Tables A7–A12 in the Online Supplemental Appendix. Results are largely similar to those presented here.

Results

Our presentation of results is anchored by our guiding research questions. First, how do the distribution and nature of adolescent romantic relationships vary within different contexts based on individual and family characteristics? To address this question, we estimate logistic regressions predicting adolescent relationship experience in each country context to show the distribution of relationships and estimate latent class analysis models for each individual country to characterize different forms of adolescent romantic relationships. Second, how does the nature of adolescent romantic relationships vary across contexts based on the nature of social and family systems? Here, we highlight differences in LCA results for relationship types across countries, returning to this variation in more depth in the conclusion.

Correlates of Adolescent Romantic Relationships

The proportion of adolescents who reported at least one romantic relationship varied across country contexts; relationship experience was most common for Mexican adolescents (55%), followed by Mozambique (37%) and Nepal (12%). (See Table A5 for additional descriptive statistics for adolescent sample, by relationship status.) Here, we present odds ratios and confidence intervals from country-specific logistic regressions modeling adolescents’ likelihood of reporting romantic relationships (Table 1).

Table 1.

Logistic regressions predicting 14-to-17-year-old adolescents’ likelihood of reporting a romantic relationship by country

Mexico Mozambique Nepal
OR (CI) OR (CI) OR (CI)
Female 1.03 0.47*** 0.38***
(0.76 – 1.40) (0.32 – 0.69) (0.25 – 0.59)
Age 1.49*** 2.03*** 1.62***
(1.29 – 1.71) (1.71 – 2.40) (1.36 – 1.93)
Primary caregiver education (v. low)
 Medium 0.95 0.99 1.44
(0.69 – 1.31) (0.63 – 1.55) (0.83 – 2.49)
 High 1.15 1.05 1.82*
(0.73 – 1.83) (0.50 – 2.22) (1.08 – 3.05)
Caste (vs. Brahmin Chetri) - -
 Hill Janajati - - 1.22
- - (0.76 – 1.97)
 Other - - 1.20
- - (0.72 – 1.99)
Household wealth index (range: −10 to 10) 1.05 0.99 0.95
(0.97 – 1.13) (0.91 – 1.08) (0.85 – 1.06)
Subjective adulthood (scale: 1–5) 1.23* 1.50*** 1.12
(1.05 – 1.45) (1.31 – 1.72) (0.90 – 1.39)
Child educational aspirations (v. low)
 Medium 1.60* 1.33 1.30
(1.10 – 2.31) (0.83 – 2.15) (0.80 – 2.12)
 High 1.25 1.70 1.01
(0.69 – 2.28) (0.79 – 3.65) (0.60 – 1.70)
Importance of marriage (v. not at all important)
 Somewhat important 1.41 2.39 0.83
(0.70 – 2.84) (0.82 – 6.93) (0.29 – 2.39)
 Neither important or unimportant 0.95 1.24 1.34
(0.48 – 1.86) (0.39 – 3.92) (0.42 – 4.32)
 Important 1.16 2.05 0.77
(0.60 – 2.24) (0.97 – 4.37) (0.28 – 2.09)
 Very important 1.77 2.34* 0.87
(0.85 – 3.71) (1.11 – 4.94) (0.28 – 2.74)
Ideal age at marriage (Range: 13–52) 0.95* 1.00 0.96
(0.90 – 1.00) (0.96 – 1.04) (0.89 – 1.03)
Desired family size 1.10 1.04 2.07**
(0.94 – 1.28) (0.92 – 1.17) (1.31 – 3.28)
Husbands should make decisions (scale: 1–5) 0.90 1.17 0.99
(0.78 – 1.04) (0.99 – 1.37) (0.80 – 1.24)
Level of caregiver supervision (scale: 5–25) 0.93** 0.95* 0.87***
(0.89 – 0.98) (0.91 – 0.99) (0.81 – 0.94)
Level of mother support (scale: 2–15) 0.99 1.07 0.99
(0.92 – 1.07) (1.00 – 1.16) (0.87 – 1.12)
Level of father support (scale: 2–15) 0.97 0.99 0.99
(0.93 – 1.02) (0.92 – 1.05) (0.90 – 1.08)
Total number of 14–17-year-olds 873 712 1201
Open in a new tab

Notes. All estimates derived from imputed data. Results are presented via odds ratios (OR) and confidence intervals (CI).

***

p < .001

**

p <.01

*

p < .05.

In Mexico, where romantic relationships are most common (Tables A1 and A5), gender differences in the experience of romantic relationships are not statistically significant. In both Mozambique and Nepal, girls are significantly less likely than boys to report having been in a relationship (p < .001). In all contexts, age is positively associated with the likelihood of reporting a romantic relationship (p < .001), as is subjectively feeling like an adult (Mexico: p = .012; Mozambique: p < .001), although this association is not statistically significant in Nepal.

Individual goals and beliefs are also related to adolescents’ reporting of relationships, but these associations vary across contexts. In Mexico, adolescents’ desired age at marriage is negatively associated with reporting a relationship (p = .035). Here, young men and women who desire to get married at older ages may postpone dating. Mexican adolescents who aspire to a post-secondary degree are more likely to report relationship experience (p = .013) versus adolescents who aspire to less than a post-secondary degree. Adolescents in Mozambique who viewed marriage as very important had a significantly higher likelihood or reporting relationship experience compared to their peers who viewed marriage as not at all important (p = .025). Adolescents in Nepal with higher desired family sizes are significantly more likely to report relationships compared to peers with lower desired family sizes (p = .002), though we note that there is little variation in desired family size among all adolescents in this context (Table A5).

Family resources, as measured by caregiver education and household wealth (and caste in Nepal), are not associated with adolescent relationship experience, except for education in Nepal, where children whose parents are highly educated are more likely to have been in a relationship compared to their peers with less-educated parents (p = .024). Across contexts, adolescents in households with more caregiver supervision were less likely to report romantic relationships (Mexico: p = .006; Mozambique: p = .010; Nepal: p < .001), but children’s perceived support by both mothers and fathers were not significantly associated with reporting a relationship.

Indicators of Class Membership

Mexico.

In Mexico (Table 2), the selected model yielded four different relationship types: physical, school-based relationships; serious, public-facing relationships; casual, school-based relationships; and casual, public-facing relationships. Physical, school-based and serious, public-facing were the most common types of relationships, accounting for 45.9% and 30.1% of relationships in the sample, respectively. These two classes also appear to be the most socially embedded types of relationship, as noted by high rates of friend, family, and parent knowledge of the relationship. Moreover, these two relationship types were the only ones where the majority of relationships had lasted at least three months. Physical intimacy, as measured by hugging and kissing, was also nearly universal in these relationship types, where item-response probabilities for these indicators ranged between 0.97 and 0.99. Serious, public-facing relationships were differentiated from physical, school-based relationships by high probabilities of adolescents having met their significant other’s family (63%), having attended a party with their partner during the relationship (85%), and the relationship lasting three or more months (94%). Casual, school-based relationships (14% of the sample) had the shortest duration and were least likely to be known to parents and family. Here, the only indicators with item-response probabilities higher than 0.6 were meeting the romantic partner at school, starting the relationship for love/companionship, and friend knowledge of the relationship. Casual, public-facing relationships (10%) were also relatively publicly visible, as indicated by parent, family, and friend knowledge of the relationship and time spent together in public rather than one-on-one. These relationships were less likely to be physically intimate than those in physical, school-based and serious, public-facing relationships.

Table 2.

Latent class prevalence and item-response probabilities for four-class model of first adolescent romantic relationships in Mexico (n = 472 relationships)

Physical, school-based Serious, public-facing Casual, school-based Casual, public-facing
Latent class prevalence
45.9% 30.1% 14.0% 10.0%
Class counts
216.6 142.0 66.3 47.1
Indicators of class membership
Met partner at school 0.61 0.55 0.74 0.59
Parents knew about relationship 0.53 0.94 0.27 0.87
Family knew about relationship 0.67 1.00 0.41 0.95
Friends knew about relationship 0.97 1.00 0.81 0.98
Met significant other’s family 0.25 0.63 0.00 0.30
Gone on trip outside community with partner 0.04 0.15 0.00 0.00
Hugged partner 0.97 1.00 0.30 0.55
Kissed partner 0.97 0.99 0.08 0.00
Possibility of marriage 0.11 0.47 0.03 0.00
Started relationship for love 0.76 0.89 0.63 0.78
Relationship lasted at least 3 months 0.58 0.94 0.23 0.54
Attended party with partner 0.38 0.85 0.06 0.31
Spent time in public square together 0.86 0.97 0.42 1.00
Open in a new tab

Notes. Latent class prevalences are based on posterior probabilities; class counts are fractional due to probabilistic assignment. Item-response probabilities indicate estimated percentage of individuals in a given class reported engaging in the behavior. Probabilities greater than or equal to 0.6 in bold to facilitate interpretation.

Mozambique.

In Mozambique (Table 3), the selected model yielded two different relationship types. The most common relationship type, with a latent class prevalence of 60.8% was physical, minimally-disclosed. These relationships appear to be relatively peer-oriented. Whereas a majority (87%) of friends knew about these relationships, parent and family knowledge of the relationship was low. Measures of physical intimacy (i.e., hugged partner, kissed partner, had sex with partner) all had high item-response probabilities. Finally, three-quarters (75%) of these relationships lasted at least three months. In contrast, 39.2% of relationships were public-facing, marriage-oriented. Adolescents who reported these relationships were very likely to have parents and family know about the relationship (61% and 84%, respectively). Moreover, 93% of these relationships lasted at least 3 months and in 88% of these cases, adolescents reported that they had had sex with their partner. Although the actual proportions differ across these two relationship types, both types of relationships have high levels of sexual intimacy.

Table 3.

Latent class prevalence and item-response probabilities for two-class model of first adolescent romantic relationships in Mozambique (n = 265 relationships)

Physical, minimally-disclosed Public-facing, marriage-oriented
Class 1 Class 2
Latent class prevalence
60.8% 39.2%
Class counts
161.1 103.9
Indicators of class membership
Met partner at school 0.35 0.38
Parents knew about relationship 0.10 0.61
Family knew about relationship 0.21 0.84
Friends knew about relationship 0.87 0.95
Met significant other’s family 0.10 0.55
Gone on trip outside community with partner 0.07 0.37
Hugged partner 0.84 0.97
Kissed partner 0.89 0.98
Possibility of marriage 0.41 0.76
Started relationship for love 0.38 0.49
Relationship lasted at least 3 months 0.75 0.93
Spent the night together 0.37 0.84
Had sex with partner 0.62 0.88
Open in a new tab

Notes. Latent class prevalences are based on posterior probabilities; class counts are fractional due to probabilistic assignment. Item-response probabilities indicate estimated percentage of individuals in a given class reported engaging in the behavior. Probabilities greater than or equal to 0.6 in bold to facilitate interpretation.

Nepal.

In Nepal (Table 4), the selected model yielded one relationship type. The small sample size (n = 142 relationships) may make it difficult to detect variation in relationship types that might be more evident if the sample were larger. It is also possible that in this context where non-marital romantic relationships are a novel form of social relationship, there is relatively little variation in experience. The relationships reported in Nepal appear to be public-facing, school-based and relatively casual. In 90% of these relationships, adolescents reported that friends knew about their relationship; in 62% of relationships, adolescents reported that they met their partners at school. Moreover, hugging (63%) but not kissing (37%) is common in relationships. Relationships were primarily started because adolescents were looking for love/companionship (53%), and a majority of these relationships (82%) lasted at least three months.

Table 4.

Latent class prevalence and item-response probabilities for one-class model of first adolescent romantic relationships in Nepal (n = 142 relationships)

Public-facing, school-based
Latent class prevalence
100.0%
Class counts
142
Indicators of class membership
Met partner at school 0.62
Parents knew about relationship 0.43
Family knew about relationship 0.51
Friends knew about relationship 0.90
Met significant other’s family 0.28
Gone on trip outside community with partner 0.44
Hugged partner 0.63
Kissed partner 0.37
Possibility of marriage 0.39
Started relationship for love 0.53
Relationship lasted at least 3 months 0.82
Spent the night together 0.09
Had sex with partner 0.06
Open in a new tab

Notes. Item-response probabilities indicate estimated percentage of individuals in a given class reported engaging in the behavior. Probabilities greater than or equal to 0.6 in bold to facilitate interpretation.

Conclusion

Adolescent romantic relationships should be examined as a part of the family formation sequence in the early life course, particularly in low- and middle-income country contexts. As the age at first marriage increases and choice marriage becomes more common (Axxe et al., 2024; Chae et al., 2021), dating practices (including adolescent relationships) will be especially important to study to better understand how, when, and in what ways young people form families (Clark & Mathur, 2012; Connolly et al., 2023; Kok et al., 2022; Manning et al., 2008). In this brief report, we have brought together developmental and sociological perspectives to better understand individual, contextual, and structural factors associated with relationship experiences in Jalisco, Mexico; Gaza province, Mozambique; and Chitwan Valley, Nepal. We have drawn on developmental perspectives via our focus on age and life course processes, as well as examining how adolescent relationships are shaped by parent-child relationships and other dynamics in the household of origin. We complement this approach with a sociological perspective that recognizes how these dynamics are also structured by broader economic, social, and family systems. We have focused on examining the individual- and household-level characteristics that are associated with adolescent first relationship experience, as well as understanding variation in relationship types within and across contexts.

In response to our guiding research questions, our analyses revealed that adolescent romantic relationships are more common (and varied) in contexts where young people are (more) involved in choosing a marital spouse, dating is appropriate and common, and where the age at marriage is increasing (Axxe et al., 2024; Chae et al., 2021; Kok et al., 2022; Manning et al., 2008). Across our study sites, relationship experience was most common in Mexico, followed by Mozambique, and lastly followed by Nepal. The correlates of romantic relationships share similarities across the three contexts. Older adolescents and those who report higher subjective adult identity are more likely to have been in a romantic relationship, consistent with developmental perspectives that see adolescent relationships as part of normative development (W. A. Collins, 2003; W. A. Collins et al., 2009; Connolly et al., 2023). In all three countries, higher levels of adult supervision are associated with lower likelihood of having been in a relationship, suggesting that higher levels of parental control may limit adolescent experiences outside the family of origin. Contrary to our expectations, household resources are not, for the most part associated with adolescent relationship experiences. Adolescent attitudes and expectations for the future are associated with having been in a relationship, but the nature of these associations varies across contexts, with educational aspirations being most salient in Mexico, importance of marriage in Mozambique, and desired family size in Nepal. These associations reflect the social meaning assigned to education, marriage, children, and family, and thus we would expect associations to be context-specific.

The types of relationships young people engage in may also be markedly different if they see these relationships as slightly more than friendship versus potential marriage partners. In Mexico, where marriage is relatively late and parents play relatively little role in choosing spouses, we found the most variation in adolescent relationships. In Nepal, where marriage is earliest and parents are highly involved in spouse selection, adolescent relationships are rare and there is limited variation in the nature of these relationships. Differences across countries make intuitive sense given that adolescent romance is most permissible (and common) in Mexico (Espinosa-Hernández & Vasilenko, 2015) and the most stigmatized in Nepal (Allendorf, 2017; Desai & Andrist, 2010). Mozambique falls in between Mexico and Nepal in terms of the acceptability of adolescent relationships and in terms of the role of parents in choosing a spouse (Axxe et al., 2024); we found that Mozambique also fell in the middle in terms of prevalence and degree of variation in relationship types. Though choice union formation and/or marriage is increasingly common throughout the world (Allendorf & Pandian, 2016; Furstenberg, 2019), this practice is certainly not uniform in the Global South. In contexts where companionate union formation is more common, dating and courtship practices and experiences, especially in adolescence, may be an increasingly important part of the family formation sequence.

The current study is not without limitations. First, all relationship data relies on self-reports from adolescents and may be subject to social desirability bias, where youth may over- or under-report relationship involvement and relationship activities (e.g., kissing, sex, etc.). Moreover, relationship reporting may also be gendered, with girls underreporting and boys overreporting relationships (Luke, Clark, et al., 2011). While we recognize this potential drawback, we take adolescent answers at face-value and call for future research to further examine relationship reports by adolescents, particularly in contexts where relationships are not normative or are stigmatized. Second, the data did not include information about the relationship quality for all countries, nor about sexual activity in Mexico. As an imperfect proxy for quality, we account for relationship length (W. A. Collins, 2003), though this likely omits other important measures of quality. While the present analysis gives us insight into different relationship types, we might see different patterns or distributions of class membership if we had access to this information. Third, questions about relationships were only asked to adolescents ages 14 to 17, so we cannot observe these associations in early adolescence. Additional qualitative research is needed to ask adolescents about what they hope to get out of specific types of relationships, how they actually define their relationships, under what circumstances they engage in specific activities, and norms about supervision. Fourth, we focused on the first relationship reported by adolescents. To the extent that the nature of relationships changes as adolescents get older or as a consequence of experiencing an earlier relationship, we may not be capturing the full range of relationships. Fifth, the data we use are cross-sectional, which precludes us from analyzing variation over time. Sixth, the data only include information on different-sex romantic relationships, and presented results may not generalize to same-sex relationships. Finally, data were collected in sites chosen to capture contexts with high levels of current or past migration, differing levels of socioeconomic development, and differing family systems. Adolescents in migrant households or communities may be exposed to different cultural or social expectations, differing levels of supervision, and qualitative and quantitative differences in dating opportunities. Future work on adolescent romance in LMICs should include non-migration contexts where possible, as well as qualitatively assess how and why migration contexts may shape relationship experiences.

Our findings highlight the importance of studying variation in adolescent romantic relationships, particularly within and across low- and middle-income country contexts, as normal developmental processes and as potential early stages of family formation sequences. We focus on how the family formation contexts of each study site is related to quantitative and qualitative relationship experience. In doing so, we heed the call for family scholars to study relationships as closely related to early adulthood outcomes and future family formation (W. A. Collins et al., 2009; Gómez-López et al., 2019; Grieger et al., 2014). Research on adolescent romantic relationships will be increasingly necessary for understanding the transition to adulthood and shifting family formation trends in LMICs, as these contexts continue to experience social and economic changes that may shape the timing and sequencing of family formation processes.

Supplementary Material

Online Supplemental Appendix

Acknowledgments:

A previous version of this paper was presented at the 2023 Annual Meeting of the Population Association of America.

Funding:

We gratefully acknowledge support from the National Institutes of Health, Eunice Kennedy Shriver Institute of Child Health and Human Development via grants P01 HD080659 (Glick, PI), 3P01HD080659-04S1 (Glick, PI; Alcaraz), OSU’s Institute for Population Research (P2C-HD058484), and PSU’s Population Research Institute (P2C-HD041025). Additional support was provided by the Department of Sociology, Ohio State University.

Footnotes

Conflicts of Interest: To the best of our knowledge, we do not have any potential conflicts of interest.

Contributor Information

Melissa Alcaraz, Brigham Young University (Department of Sociology), Provo, UT, USA.

Sarah R. Hayford, Ohio State University (Department of Sociology), Columbus, OH, USA

REFERENCES

  1. Ahearn LM (2001). Invitations to Love: Literacy, Love Letters, and Social Change in Nepal. Univ of Michigan Pr. [Google Scholar]
  2. Alcaraz M, Hayford SR, & Glick JE (2022). Desired fertility and educational aspirations: Adolescent goals in rapidly changing social contexts. Journal of Marriage and Family, 84(1), 7–31. 10.1111/jomf.12815 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Allendorf K (2017). Conflict and Compatibility? Developmental Idealism and Gendered Differences in Marital Choice. Journal of Marriage and Family, 79(2), 337–355. 10.1111/jomf.12339 [DOI] [Google Scholar]
  4. Allendorf K, & Pandian RK (2016). The Decline of Arranged Marriage? Marital Change and Continuity in India. Population and Development Review, 42(3), 435–464. 10.1111/j.1728-4457.2016.00149.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Aloisio KM, Micali N, Swanson SA, Field A, & Horton NJ (2014). Analysis of partially observed clustered data using generalized estimating equations and multiple imputation. The Stata Journal: Promoting Communications on Statistics and Stata, 14(4), 863–883. 10.1177/1536867X1401400410 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Ankomah A (1999). Sex, Love, Money and AIDS: The Dynamics of Premarital Sexual Relationships in Ghana. Sexualities, 2(3), 291–308. 10.1177/136346099002003002 [DOI] [Google Scholar]
  7. Archambault JS (2017). Mobile Secrets: Youth, Intimacy, and the Politics of Pretense in Mozambique. The University of Chicago Press. [Google Scholar]
  8. Auslander BA, Short MB, Succop PA, & Rosenthal SL (2009). Associations Between Parenting Behaviors and Adolescent Romantic Relationships. Journal of Adolescent Health, 45(1), 98–101. 10.1016/j.jadohealth.2008.12.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Axxe E, Hayford SR, & Eggum ND (2022). Youth’s family and non-family roles as predictors of subjective adulthood in three low-income agricultural settings. Journal of Research on Adolescence, Early View. 10.1111/jora.12731 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Axxe E, Hayford SR, & Ghimire DJ (2024). Household Migration and Adolescent Marriage Choice in Settings with Changing Marriage Systems1. Journal of Comparative Family Studies, 55(1), 4–31. (181198899). 10.3138/jcfs.55.1.02 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Batyra E, Kohler H-P, & Furstenberg FF (2021). Changing Gender Gaps in the Timing of First Union Formation and Sexual Initiation in Sub-Saharan Africa. Population and Development Review, 47(2), 289–322. 10.1111/padr.12405 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Bingenheimer JB, Roche KM, & Blake SM (2017). Family Adult Awareness of Adolescents’ Premarital Romantic and Sexual Relationships in Ghana. Youth & Society, 49(3), 341–368. 10.1177/0044118X15581168 [DOI] [Google Scholar]
  13. Bongaarts J, Mensch BS, & Blanc AK (2017). Trends in the age at reproductive transitions in the developing world: The role of education. Population Studies, 71(2), 139–154. 10.1080/00324728.2017.1291986 [DOI] [PubMed] [Google Scholar]
  14. Campero L, Walker D, Atienzo EE, & Gutierrez JP (2011). A quasi-experimental evaluation of parents as sexual health educators resulting in delayed sexual initiation and increased access to condoms. Journal of Adolescence, 34(2), 215–223. 10.1016/j.adolescence.2010.05.010 [DOI] [PubMed] [Google Scholar]
  15. Cavanagh SE (2007). The Social Construction of Romantic Relationships in Adolescence: Examining the Role of Peer Networks, Gender, and Race*. Sociological Inquiry, 77(4), 572–600. 10.1111/j.1475-682X.2007.00207.x [DOI] [Google Scholar]
  16. Cavanagh SE, Crissey SR, & Raley RK (2008). Family Structure History and Adolescent Romance. Journal of Marriage and Family, 70(3), 698–714. 10.1111/j.1741-3737.2008.00515.x [DOI] [Google Scholar]
  17. Chae S, Agadjanian V, & Hayford SR (2021). Bridewealth marriage in the 21st century: A case study from rural Mozambique. Journal of Marriage and Family, 83(2), 409–427. 10.1111/jomf.12725 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Clark S, & Agnant K (2025). Charting New Courses to Adulthood in the Global South. Population and Development Review, 51(1), 181–211. 10.1111/padr.12711 [DOI] [Google Scholar]
  19. Clark S, Kabiru C, & Mathur R (2010). Relationship Transitions Among Youth in Urban Kenya. Journal of Marriage and Family, 72(1), 73–88. 10.1111/j.1741-3737.2009.00684.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Clark S, & Mathur R (2012). Dating, Sex, and Schooling in Urban Kenya. Studies in Family Planning, 43(3), 161–174. 10.1111/j.1728-4465.2012.00315.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Collins LM, & Lanza ST (2010). Latent Class and Latent Transition Analysis: With Applications in the Social, Behavioral, and Health Sciences. Wiley. [Google Scholar]
  22. Collins WA (2003). More than Myth: The Developmental Significance of Romantic Relationships During Adolescence. Journal of Research on Adolescence, 13(1), 1–24. 10.1111/1532-7795.1301001 [DOI] [Google Scholar]
  23. Collins WA, Welsh DP, & Furman W (2009). Adolescent Romantic Relationships. Annual Review of Psychology, 60(1), 631–652. 10.1146/annurev.psych.60.110707.163459 [DOI] [PubMed] [Google Scholar]
  24. Connolly J, Shulman S, & Benvenuto K (2023). Romantic relationships in adolescence and early adulthood. In Crockett LJ, Carlo G, & Schulenberg JE (Eds.), APA handbook of adolescent and young adult development. (pp. 243–258). American Psychological Association. 10.1037/0000298-015 [DOI] [Google Scholar]
  25. Corsi DJ, Neuman M, Finlay JE, & Subramanian S (2012). Demographic and health surveys: A profile. International Journal of Epidemiology, 41(6), 1602–1613. 10.1093/ije/dys184 [DOI] [PubMed] [Google Scholar]
  26. Desai S, & Andrist L (2010). Gender scripts and age at marriage in India. Demography, 47(3), 667–687. 10.1353/dem.0.0118 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Elder GH (1998). The life course as developmental theory. Child Development, 69(1), 1–12. 10.1111/j.1467-8624.1998.tb06128.x [DOI] [PubMed] [Google Scholar]
  28. Espinosa-Hernández G, & Vasilenko SA (2015). Patterns of relationship and sexual behaviors in Mexican adolescents and associations with well-being: A latent class approach. Journal of Adolescence, 44, 280–290. 10.1016/j.adolescence.2015.08.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Filmer D, & Pritchett LH (2001). Estimating wealth effects without expenditure data-or tears: An application to educational enrollments in states of India. Demography, 38(1), 115–132. JSTOR. 10.2307/3088292 [DOI] [PubMed] [Google Scholar]
  30. Furman W, Collibee C, Lantagne A, & Golden RL (2019). Making Movies Instead of Taking Snapshots: Studying Change in Youth’s Romantic Relationships. Child Development Perspectives, 13(3), 135–140. 10.1111/cdep.12325 [DOI] [Google Scholar]
  31. Furman W, & Shaffer L (2003). The role of romantic relationships in adolescent development. Adolescent Romantic Relations and Sexual Behavior: Theory, Research, and Practical Implications, 3–22. [Google Scholar]
  32. Furstenberg FF (2019). Family Change in Global Perspective: How and Why Family Systems Change. Family Relations, 68(3), 326–341. 10.1111/fare.12361 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Gaalen RIV, & Dykstra PA (2006). Solidarity and Conflict Between Adult Children and Parents: A Latent Class Analysis. Journal of Marriage and Family, 68(4), 947–960. 10.1111/j.1741-3737.2006.00306.x [DOI] [Google Scholar]
  34. Gibbons JL, & Poelker KE (2019). Adolescent development in a cross-cultural perspective. In Keith KD (Ed.), Cross-Cultural Psychology (pp. 190–215). John Wiley & Sons, Ltd. 10.1002/9781119519348.ch9 [DOI] [Google Scholar]
  35. Giordano PC (2003). Relationships in Adolescence. Annual Review of Sociology, 29, 257–281. [Google Scholar]
  36. Giordano PC, Manning WD, & Longmore MA (2005). The Romantic Relationships of African-American and White Adolescents. The Sociological Quarterly, 46(3), 545–568. 10.1111/j.1533-8525.2005.00026.x [DOI] [Google Scholar]
  37. Gómez-López M, Viejo C, & Ortega-Ruiz R (2019). Well-Being and Romantic Relationships: A Systematic Review in Adolescence and Emerging Adulthood. International Journal of Environmental Research and Public Health, 16(13), Article 13. 10.3390/ijerph16132415 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Gowen LK, Catania JA, Dolcini MM, & Harper GW (2014). The Meaning of Respect in Romantic Relationships Among Low-Income African American Adolescents. Journal of Adolescent Research, 29(5), 639–662. 10.1177/0743558414528978 [DOI] [Google Scholar]
  39. Grant MJ (2012). Girls’ schooling and the perceived threat of adolescent sexual activity in rural Malawi. Culture, Health & Sexuality, 14(1/2), 73–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Grant MJ, Lloyd C, & Mensch B (2013). Menstruation and School Absenteeism: Evidence from Rural Malawi. Comparative Education Review, 57(2), 260–284. 10.1086/669121 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Grieger L, Kusunoki Y, & Harding DJ (2014). The social contexts of adolescent romantic relationships. Focus, 31(1), 3. [Google Scholar]
  42. Hagenaars JA, & Halman LC (1989). Searching for ideal types: The potentialities of latent class analysis. European Sociological Review, 5(1), 81–96. 10.1093/oxfordjournals.esr.a036509 [DOI] [Google Scholar]
  43. Halpern-Meekin S, & Turney K (2022). Family Instability and Adolescent Relationships: The Role of Parental Relationship Churning. Youth & Society, 54(5), 730–765. 10.1177/0044118X20985148 [DOI] [Google Scholar]
  44. Heuveline P, & Timberlake JM (2004). The Role of Cohabitation in Family Formation: The United States in Comparative Perspective. Journal of Marriage and Family, 66(5), 1214–1230. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. INEGI. (2024a). Estadística de Matrimonios (EMAT) 2023. INEGI. https://www.inegi.org.mx/contenidos/saladeprensa/boletines/2024/EMAT/EMAT2023.pdf [Google Scholar]
  46. INEGI. (2024b). Usa la estadística y la geografía para descubrir México. Cuéntame de México. https://cuentame.inegi.org.mx/explora/poblacion/nupcialidad/ [Google Scholar]
  47. Instituto Nacional de Estatística (INE) e ICF. (2024). Inquérito Demográfico e de Saúde em Moçambique 2022–23. INE e ICF. https://dhsprogram.com/pubs/pdf/FR389/FR389.pdf [Google Scholar]
  48. Johnson MK, Crosnoe R, & Elder GH Jr. (2011). Insights on adolescence from a life course perspective. Journal of Research on Adolescence, 21(1), 273–280. 10.1111/j.1532-7795.2010.00728.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Juárez F, & Gayet C (2014). Transitions to adulthood in developing countries. Annual Review of Sociology, 40(1), 521–538. 10.1146/annurev-soc-052914-085540 [DOI] [Google Scholar]
  50. Kansky J, & Allen JP (2018). Long-Term Risks and Possible Benefits Associated with Late Adolescent Romantic Relationship Quality. Journal of Youth and Adolescence, 47(7), 1531–1544. 10.1007/s10964-018-0813-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Kok MC, van Eldik Z, Kakal T, Munthali A, Menon JA, Pires P, Baatsen P, & van der Kwaak A (2022). Being dragged into adulthood? Young people’s agency concerning sex, relationships and marriage in Malawi, Mozambique and Zambia. Culture, Health & Sexuality, 24(6), 767–781. 10.1080/13691058.2021.1881618 [DOI] [PubMed] [Google Scholar]
  52. Laborde ND, vanDommelen-Gonzalez E, & Minnis AM (2014). Trust – that’s a big one: Intimate partnership values among urban Latino youth. Culture, Health & Sexuality, 16(9), 1009–1022. 10.1080/13691058.2014.921837 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Lantagne A, & Furman W (2017). Romantic relationship development: The interplay between age and relationship length. Developmental Psychology, 53(9), 1738–1749. 10.1037/dev0000363 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Luke N, Clark S, & Zulu EM (2011). The Relationship History Calendar: Improving the Scope and Quality of Data on Youth Sexual Behavior. Demography, 48(3), 1151–1176. 10.1007/s13524-011-0051-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Luke N, Goldberg RE, Mberu BU, & Zulu EM (2011). Social Exchange and Sexual Behavior in Young Women’s Premarital Relationships in Kenya. Journal of Marriage and Family, 73(5), 1048–1064. 10.1111/j.1741-3737.2011.00863.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Manning WD, Giordano PC, & Longmore MA (2008). Adolescent Dating Relationships: Implications for Understanding Adult Intimate Unions. [Google Scholar]
  57. Marphatia AA, Saville NM, Amable GS, Manandhar DS, Cortina-Borja M, Wells JC, & Reid AM (2020). How Much Education Is Needed to Delay Women’s Age at Marriage and First Pregnancy? Frontiers in Public Health, 7. https://www.frontiersin.org/articles/10.3389/fpubh.2019.00396 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Meier A, & Allen G (2009). Romantic Relationships from Adolescence to Young Adulthood: Evidence from the National Longitudinal Study of Adolescent Health. The Sociological Quarterly, 50(2), 308–335. 10.1111/j.1533-8525.2009.01142.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Meinster MO, & Rose KC (2001). Longitudinal Influences of Educational Aspirations and Romantic Relationships on Adolescent Women’s Vocational Interests. Journal of Vocational Behavior, 58(3), 313–327. 10.1006/jvbe.2000.1772 [DOI] [Google Scholar]
  60. Ministerio da Saude (MISAU), Instituto Nacional de Estatística (INE), & ICF International (ICFI). (2013). Moçambique Inquérito Demográfico e de Saúde 2011. MISAU, INE e ICFI. https://dhsprogram.com/pubs/pdf/fr266/fr266.pdf [Google Scholar]
  61. Ministry of Health and Population [Nepal], New ERA, & ICF. (2023). Nepal Demographic and Health Survey 2022. The DHS Program. https://dhsprogram.com/pubs/pdf/FR379/FR379.pdf [Google Scholar]
  62. National Research Council and Institute of Medicine. (2004). Growing Up Global: The Changing Transitions to Adulthood in Developing Countries. 10.17226/11174 [DOI] [Google Scholar]
  63. Olmstead SB (2020). A Decade Review of Sex and Partnering in Adolescence and Young Adulthood. Journal of Marriage and Family, 82(2), 769–795. 10.1111/jomf.12670 [DOI] [Google Scholar]
  64. Ortíz-Ávila E (2020). Sociodemographic strata and patterns related to the first union and first motherhood in Mexico, 2018. Población y Salud En Mesoamérica, 18(1), 205–226. 10.15517/psm.v18i1.40350 [DOI] [Google Scholar]
  65. Parkes J, Heslop J, Januario F, Oando S, & Sabaa S (2016). Between tradition and modernity: Girls’ talk about sexual relationships and violence in Kenya, Ghana and Mozambique. Comparative Education, 52(2), 157–176. 10.1080/03050068.2016.1142741 [DOI] [Google Scholar]
  66. Poulin M (2007). Sex, money, and premarital partnerships in southern Malawi. Social Science & Medicine, Informed Consent in a Changing Environment, 65(11), 2383–2393. 10.1016/j.socscimed.2007.05.030 [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Royston P, & White IR (2011). Multiple imputation by chained equations (MICE): Implementation in Stata. Journal of Statistical Software, 45(1), 1–20. 10.18637/jss.v045.i04 [DOI] [Google Scholar]
  68. Schneider D, & Hastings OP (2015). Socioeconomic Variation in the Effect of Economic Conditions on Marriage and Nonmarital Fertility in the United States: Evidence From the Great Recession. Demography, 52(6), 1893–1915. 10.1007/s13524-015-0437-7 [DOI] [PubMed] [Google Scholar]
  69. Settersten RA Jr, Furstenberg FF, & Rumbaut RG (Eds.). (2008). On the Frontier of Adulthood: Theory, Research, and Public Policy. University of Chicago Press. https://press.uchicago.edu/ucp/books/book/chicago/O/bo3533908.html [Google Scholar]
  70. Sheppard P, Pearce MS, & Sear R (2016). How does childhood socioeconomic hardship affect reproductive strategy? Pathways of development. American Journal of Human Biology, 28(3), 356–363. 10.1002/ajhb.22793 [DOI] [PMC free article] [PubMed] [Google Scholar]
  71. Shulman S, & Connolly J (2013). The Challenge of Romantic Relationships in Emerging Adulthood: Reconceptualization of the Field. Emerging Adulthood, 1(1), 27–39. 10.1177/2167696812467330 [DOI] [Google Scholar]
  72. Vargas Trujillo E, & Barrera F (2002). Adolescencia, relaciones románticas y actividad sexual: Una revisión. Revista Colombiana de Psicología, 11, 115–134. [Google Scholar]
  73. Villalta SI, Goldberg RE, & Tienda M (2023). Adolescent Partnership Quality and Emotional Health: Insights from an Intensive Longitudinal Study. Social Forces, 101(3), 1422–1459. 10.1093/sf/soac043 [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. World Bank. (2025). Mean age at first marriage. World Bank Gender Data Portal. https://genderdata.worldbank.org/en/indicator/sp-dyn-smam [Google Scholar]
  75. Wurpts IC, & Geiser C (2014). Is adding more indicators to a latent class analysis beneficial or detrimental? Results of a Monte-Carlo study. Frontiers in Psychology, 5. 10.3389/fpsyg.2014.00920 [DOI] [PMC free article] [PubMed] [Google Scholar]
  76. Xia M, Fosco GM, Lippold MA, & Feinberg ME (2018). A Developmental Perspective on Young Adult Romantic Relationships: Examining Family and Individual Factors in Adolescence. Journal of Youth and Adolescence, 47(7), 1499–1516. 10.1007/s10964-018-0815-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  77. Zimmer-Gembeck MJ, Petegem SV, Ducat W, Clear S, & Mastro S (2017). Parent–adolescent relationships and romantic development: A review and argument for research on autonomy-supportive parenting. In Autonomy in Adolescent Development. Psychology Press. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Online Supplemental Appendix
NIHMS2204071-supplement-Online_Supplemental_Appendix.docx (65.1KB, docx)

RESOURCES