A. Study population

Many populations in Nepal traditionally practise arranged marriage. However, over the past 50 years Nepal has begun a dramatic shift in marriage practices, including fewer child and adolescent marriages, more inter-caste marriages and greater offspring participation in spouse choice (Ghimire, Axinn, Yabiku, & Thornton, Reference Ghimire, Axinn, Yabiku and Thornton2006). This shift has produced a cohort of individuals in Nepal whose marriages vary from arranged to non-arranged. In addition to this significant variation in marriage type, the national divorce rate is less than 1% and remarriage, especially for women, is uncommon (Nepal Population and Housing Census 2011, 2012). Thus, choosing a spouse is effectively permanent, making this a useful context in which to examine the effects of different modes of spouse choice.

In many castes and ethnic groups in Nepal, marriages are traditionally restricted to members of the same caste and subcaste, but not within the same gotra (lineage). Caste is primarily inherited through males, but there are also subcastes that are specific to intermarried couples that can change the caste of their children. For example, the Khatri Chhetri subcaste denotes the presence of an intermarriage between Brahmin man and non-Brahmin woman (Höfer, Reference Höfer1979). Parents may informally recruit a lami (matchmaker), usually extended kin or acquaintances who live in other locations, to help search for suitable spouses. After marriage, women traditionally live with their husband and his family. Historically, dowry was an important part of marriage negotiations, providing financial transfer from the bride's family to the groom and his family. While the practice has been illegal in Nepal since 2009, the law primarily prevents the groom's family from requesting specific items rather than preventing significant marital gift giving. While daughters traditionally receive financial transfers at marriage, sons primarily receive financial transfers through inheritances. Prior to 2017, Nepali law required inheritances to be split equally among all sons but did not require any transfer to daughters (National Civil Code Act, 2017). Nepal does not legally allow marriage until age 20 for either men or women (National Civil Code Act, 2017). Most marriages adhere to this law, but it is not always well-enforced, especially in rural areas where marriages are not always registered with a government office.

The data for this study were collected in a large community (over 20,000 people) in the Dhading District of Nepal, which lies directly west of the Kathmandu Valley. The community has access to a major highway, and socioeconomic status varies with proximity to the highway. Most individuals in the community are farmers or shopkeepers, and women traditionally work within the household. Many people under 30 years of age have achieved the equivalent of a high-school education or beyond, but education levels decrease with age and for women. The community includes members of a variety of castes and ethnic groups, with the highest representation coming from Brahmins, Chhetris, Tamangs and Newars. While most marriages in the community adhere to the national law about marriage age, some people in the community, particularly women, marry as teenagers. Marriage under age 16 is rare among individuals married in the last 40 years. Based on data we collected in the community, 78.6% of people in the oldest generation have said their spouse was primarily chosen by their parents. Of people married in the last 15 years, only 56.2% said their parents were primarily responsible for their spouse choice, mirroring the widespread shift in Nepal from arranged to non-arranged marriages. The variation in marriage type within a single generation makes this community a good place to examine the diversity of methods of spouse choice and explore their potential fitness effects (Figure 1).

Figure 1. Image of the study site in Dhading, Nepal, showing the highway that passes through the town, terraced fields and hilly terrain.

B. Data collection

Semi-structured focus group discussions were conducted in 2019 and 2022. Participants were recruited via convenience sampling in various neighbourhoods within the community, and recruitment was done at different times of day to target people with varying work schedules. Participants were told they could bring a friend or neighbour of the same gender and approximate age to participate in the group as well, in the hopes that some familiarity within the group would facilitate more casual and lively conversations. A few (three or four) women across these groups brought mothers- or daughters-in-law to participate; in these cases, they were placed into separate groups based on age. All participants were recruited by local research assistants, as previous experience showed increased trust and willingness to participate when local assistants were involved. Men and women were recruited from three life stages that roughly mirror age cohorts: never-married adults ages 18–30, adults married within the last 10 years and married adults with adult children (hereafter referred to as unmarried, recently married and older, respectively).

Individual focus groups comprised homogeneous gender and age groups (e.g. older women). Discussions were facilitated in Nepali by a trained local research assistant of the same gender as the participants and were limited to 1 hour. In total, we conducted 19 focus groups with an average size of 5.6 participants. The number and composition of focus groups can be found in Table 1. Most of the unmarried individuals were students at the time of the discussions (50%). A majority of the married men ran local businesses (52%) or were in the service industry (33%). Most women interviewed were either farmers (60%) or shopkeepers (33%). While participant incomes were not collected, generally those who work in business or run shops tend to have higher incomes than farmers or those who do service jobs.

Table 1. Focus group summary statistics for each demographic category

Note: ‘Unmarried’ includes adults aged 18–30 who have never married, ‘Recently married’ includes individuals of any age married within the last 10 years, and ‘Older’ includes ever-married adults with children aged 16 or older. All ages are in years.

Focus groups were designed to target topics that were relevant to each group's life stage. The facilitators were given a list of potential questions and vignettes to generate discussion but were instructed to allow conversation to flow naturally and alter the questions or their order accordingly. Unmarried adults were asked questions about dating, the marriage process, desirable qualities of potential spouses and relationships with their parents. Recently married adults were asked about the process of finding a spouse, their lives after marriage, and about starting a family. Older adults were asked questions relating to their children's marriages (or future marriages) and their own lives after marriage. Many discussion topics were chosen to target our specific research questions:

Most of the questions prompted participants to think about the community generally rather than about their own experiences to increase openness, especially surrounding more difficult or taboo topics, such as dowry, marital disagreements, affairs and fertility.

All focus groups were recorded, transcribed in Nepali, and translated into English by a bilingual native Nepali speaker. The Nepali–English translations were checked by a second bilingual native Nepali speaker (Author P.U.) to ensure their accuracy. Translations were also checked to make sure that important context was not lost, paying close attention to culturally specific words or phrases that have implied meanings or do not translate literally. Where necessary for clarity, notes on such meanings are inserted into quotes in square brackets.

C. Data analysis

English transcripts were read and independently coded for content by S.C. and A.W. When a new topic was identified, it was added to a codebook and given a brief description to maintain consistency of use in future coding. The coders then cross-checked each other's work for accuracy and comprehensiveness. All authors read the transcripts in full, and each author independently kept notes on themes they noticed, statements that were surprising and clear differences between demographic groups.

After this initial coding, statements were reorganised by code with tags denoting the gender/age group where it originated. Following the reorganisation, all statements from each theme were read together. The authors noted any differences based on age or gender, as well as other distinctions drawn by participants, including statements about different behaviours among the rich/poor, rural/urban or past/present. These themes and patterns were discussed in biweekly meetings in which the authors compared their observations with each other's observations, as well as to their own notes from earlier readings of the transcripts. The patterns and themes described in this analysis were agreed upon by all authors.

A. Is the mode of spouse choice a dichotomous variable?

Focus group discussions identified three types of spouse choice in marriage: arranged marriages, love marriages and elopements. Each vary on the autonomy of the spouse-choice decision (summarised in Table 2).

Table 2. Summary of marriage types and their associated forms of support

B. Do parents and offspring agree on the ideal qualities of an in-law or spouse?

The focus groups were asked to describe what qualities make a good spouse, or what qualities they would look for in a spouse for a son or daughter. Respondents freely listed 26 different qualities. A brief description of the data across all desirable qualities is included in Supplementary Table S1, but only traits that were expected to produce disagreement based on previous research on parent–offspring disagreement over spouse choice (Agey et al., Reference Agey, Morris, Chandy and Gaulin2021; Apostolou, Reference Apostolou2010a; Buunk et al., Reference Buunk, Park and Dubbs2008; Perilloux et al., Reference Perilloux, Fleischman and Buss2011) are discussed below. The results are summarised in Figure 2.

Figure 2. Potential for parent–offspring agreement and disagreement over traits in a potential spouse. These diagrams represent the potential for disagreement between daughters and their parents (a) and sons and their parents (b) based on focus group discussions with unmarried and older participants. Areas of overlap represent agreement on the listed traits, while non-overlapping areas represent disagreement. (a) Daughters, mothers and fathers agree that a potential spouse should be of a similar caste and come from a family with a good reputation. They also agree that marriage should occur in the daughter's mid-20s and not occur in their teenage years. Mothers and daughters agree that physical attractiveness is desirable and important in a potential spouse. Daughters sometimes express a preference to marry after age 30, which would not be desirable for mothers or fathers. (b) Sons, mothers and fathers agree that a potential spouse should be of a similar caste and come from a family with a good reputation. They should also agree that marriage should occur in the son's mid-20s or 30s and not occur in their teenage years. Sons and fathers agree that physical attractiveness should not be a priority for choosing a spouse, while mothers think that physical attractiveness is desirable and important.

C. What kinds of social and financial effects are tied to spouse choice decisions?

It is possible that allowing parents to choose a spouse, or at least seeking their approval for a spouse, is tied to certain benefits for the couple that would be lost if parents are not involved in the spouse choice. The results below are summarised in Table 2.

A. Spouse choice is not a dichotomous variable

It is clear from these discussions that arranged and non-arranged marriage do not represent a dichotomy. Participants identified three types of marriages but described variation within each type. Arranged marriage may or may not have offspring approval of the parents’ choice, implying that a broad category for ‘arranged marriage’ may obscure important variation in the degree of choice offspring have. Both love marriage and elopement would technically be considered non-arranged marriages, but love marriage and elopement have very different levels of parental involvement before and after marriage. Thus, sorting marriages based on who initially found the spouse may still not capture the full range of variation in spouse choice decisions. In Dhading, spouse choice may be best measured as an index of several variables, including who originally found the spouse, who initially suggested marriage to that person, whether permission was sought and granted from other parties prior to marriage, and the perceived degree to which each party was able to influence the other in their choice. Furthermore, information offered by participants suggests that husbands and wives may answer questions about spouse choice differently. Approval for a marriage could come from the bride's family, the groom's family, or from both families, and each of these situations could have different implications for the couple's post-marital life. Failing to capture these sources of variation may make it difficult to accurately differentiate the consequences of partner choice.

B. Parents and offspring often agree over the ideal qualities of a spouse

Older respondents were more likely than unmarried respondents to name caste as the first quality they would consider in a spouse for their children. While this may indicate that parents care more about caste than unmarried individuals, there was more willingness to compromise on caste-matching than expected. Parents may face a difficult choice when their children want to marry someone of a different caste because, if parents do not approve of an inter-caste match, the couple may elope instead. Parents may thus be willing to approve an inter-caste love marriage if it produces less stigma than an inter-caste elopement.

While not discussed by participants in these focus groups, the level of compromise over caste probably varies with the size of the gap between the two castes or ethnicities (which are both integrated into a singular system in Nepal). In Nepal, the phrase pani chalne indicates castes and ethnicities designated as ‘pure’ enough to drink water given by them. The lower disadvantaged (i.e. Dalit) castes are excluded from this category, but other ethnic groups are considered pani chalne (Höfer, Reference Höfer1979). In informal conversation in this community, we have heard parents say they would accept any marriage from a pani chalne group. Therefore, it is likely that a marriage between the most advantaged caste levels would be more acceptable than an inter-ethnic marriage between two pani chalne groups, and both of those matches would be more acceptable than a marriage where only one party is in a disadvantaged caste. Because caste is passed primarily through males and hypergamy is formally permitted (Höfer, Reference Höfer1979), parents may be more accepting of sons marrying a lower-caste woman than a daughter marrying a lower-caste man. Thus, women probably face greater sanctions than men for marrying below their caste.

Despite physical attractiveness producing disagreement between parents and offspring in other studies (Agey et al., Reference Agey, Morris, Chandy and Gaulin2021; Apostolou, Reference Apostolou2008a; Bovet et al., Reference Bovet, Raiber, Ren, Wang and Seabright2018; Buunk et al., Reference Buunk, Park and Dubbs2008; Perilloux et al., Reference Perilloux, Fleischman and Buss2011), these discussions do not indicate strong conflict over this quality. For unmarried men, attractiveness seemed to be a threshold variable: as long as a minimum standard was met, they could be satisfied. The desire for a beautiful daughter-in-law that many older women voiced may mean that sons are able to easily achieve this threshold, even in arranged marriages. Younger women also expressed a desire for physically attractive husbands, indicating that women and their mothers agree in this community. However, older men explicitly stated that a potential husband does not need to be handsome as long as he can provide; thus, disagreement over physical attractiveness is more likely between fathers and daughters than between daughters and mothers or sons and either parent in this community.

Older individuals appeared slightly more attentive than unmarried individuals to the quality of the potential spouse's family. However, unmarried people, especially women, were very clear that the behaviour of the intended spouse, especially around drinking or affairs, was more important than the quality of the family; for the marrying individuals a bad apple from a good tree was not an acceptable partner. Parents are probably more attentive to the social network and prestige benefits of arranging matches for their children, and they may sometimes perceive this as being more important than the quality of the potential spouse themselves. Parents might also be worse than offspring at obtaining truthful information about the behaviour of a potential spouse (e.g. because such behaviours may be hidden from older individuals in the community). Where there is discordance between the quality of the family and the quality of the potential spouse, there will probably be more disagreement between parents and offspring. However, in cases where the potential spouse's behaviour is acceptable, then an arranged marriage would satisfy the preferences of both parents and offspring.

There may also be disagreement over marriage timing, especially between parents and daughters. Parents could be pushing daughters to marry earlier than sons for several reasons – female reproduction is more closely tied to age, dowry is also traditionally lower for younger brides, and parents can also exert more control over spouse choice when offspring are younger. While older individuals expressed a lot of concern about very young people eloping, they did not seem as concerned about marriage age in arranged marriages. This indicates that parents may be hesitant to allow love marriages to proceed if they consider their children to be too young or immature to choose their own partner. Parents may reject love matches when their offspring are under the parents’ ideal marriage age. Conversely, young women facing strong parental pressure to have an arranged marriage before they are ready may have to choose between agreeing to their parents’ choice or eloping with a man of their choosing. These factors may explain why elopement is often done by young people. This pattern mirrors patterns of early marriage in Tanzania, in which parents sometimes arrange early marriages to secure socioeconomic stability for their daughters while at the same time being concerned about non-arranged early marriages that appear unstable or based on poor decision making (Baraka, Lawson, Schaffnit, Wamoyi, & Urassa, Reference Baraka, Lawson, Schaffnit, Wamoyi and Urassa2022). Many young women in Dhading expressed a desire to finish school, build a career, and become more financially independent before marriage. They may see these activities as a way to increase their own status, which could be converted to better matches (Schaffnit et al., Reference Schaffnit, Page, Lynch, Spake, Sear, Sosis and Shenk2022), more extensive kin networks (Shenk, Towner, Voss, & Alam, Reference Shenk, Towner, Voss and Alam2016), more parental investment and better outcomes for their children (Alami et al., Reference Alami, von Rueden, Seabright, Kraft, Blackwell, Stieglitz and Gurven2020). Because, owing to legal changes and societal trends, marriage age has been steadily increasing in this community (an offspring preference), this could indicate that more compromise between parents and their children over marriage age may be occurring in recent years.

The overall picture shows some avenues for parent–offspring disagreement, but also shows many opportunities for cooperation and compromise. Parents and their children may work together to meet minimum standards on physical attractiveness, quality of behaviour and family background. Parents may compromise on inter-caste marriages in cases where castes are similar or where elopement is a risk. While disagreement about marriage age may also be common, the levels may be decreasing as average marriage age rises in Nepal. Thus, disagreement may only be present in some parent–offspring dyads, and the risk of disagreement over particular qualities is probably asymmetrical for sons and daughters. Likewise, individuals who are shy or do not have their own prospects may obtain better mates through arranged marriage than if they choose on their own, producing very little disagreement. Similarly, individuals who can independently obtain high-quality mates may be able to meet both theirs and their parents’ priorities, again producing very little disagreement. Because elopements (which indicate disagreement) were consistently reported as rare in discussions, it could be the case that parents and offspring are often able to agree on spouse choice in this community. Thus, in cases where both parties cooperate to choose a spouse (i.e. when parents choose a spouse and the offspring approve or when offspring choose a spouse and parents approve), there may be few fitness consequences.

C. Arranged marriages receive compensatory benefits

Generally, those who choose their spouse with no parental input seem to have less access to family support, at least immediately following their marriages. The consequences of eloping appear stronger for women than for men, as many respondents directly stated that daughters-in-law face more discrimination for eloping than sons-in-law. Even if parents later accept an elopement, it is clear that there is less support if the marriage is unsuccessful or faces hardships, as many participants noted that family cannot get involved in conflicts between self-selected spouses. This decreased familial involvement may have significant ramifications in instances of abuse and intimate partner violence, particularly for women, whose statements in these discussions indicated their desire for strong family support. Dowry, which is unlikely to be paid in elopements, was associated with better relationships with in-laws after marriage, consistent with other studies examining relationships between daughters- and mothers-in-law in Nepal (Diamond-Smith, Dahal, Puri, & Weiser, Reference Diamond-Smith, Dahal, Puri and Weiser2020). This could be because marital transfers can help legitimise new couples in the eyes of the family and community, as bridewealth appears to do in Ghana (Akurugu, Dery, & Domanban, Reference Akurugu, Dery and Domanban2022). The lack of dowry could therefore produce more conflict, additional stress, or poorer access to resources in the marital household if the couple lives with the husband's family. Respondents noted that husbands were more likely to take their wife's side during family disagreements in eloped and love marriage couples. While this could attenuate some of the stress and improve treatment of a new wife in her husband's family home, this situation was also tied to the couple establishing a separate nuclear household, thereby further distancing themselves from family. From these lines of evidence, women who choose their own spouses may suffer in terms of support from both her own family and her husband's family. This reduction in material support or increase in interpersonal conflict could increase undernutrition and psychosocial stress, affecting health, pregnancy and birth outcomes (Hobel, Goldstein, & Barrett, Reference Hobel, Goldstein and Barrett2008; Schneiderman, Ironson, & Siegel, Reference Schneiderman, Ironson and Siegel2005). Since families are more likely to accept an eloping couple following the birth of a child, the effect of reduced familial support may be the strongest on first births.

Arranged couples, who regularly receive gifts/dowries, may be able to achieve financial stability more often or more quickly than couples that eloped, who are unlikely to receive dowries or gifts at marriage. Marital gifts contingent on approval from the bride's family can contribute to the financial stability of the new couple, especially when they consist of cash, land or vehicles that can increase access to jobs and markets. The relationship between marriage type and financial support is less clear for men in this community. While men have always been legally guaranteed to receive an inheritance from their parents, parents may distribute inheritances at different times. If the timing of inheritance is tied to marriage type, that opens an avenue through which resources could be affected by marriage type for men in this community.

Eloping couples were expected to be more financially self-sufficient, and their parents’ acceptance of the marriage was often tied to the couple's financial independence. If eloping couples have less assistance financially from the bride's family (e.g. less gift-giving/dowry) or a delay in assistance from the groom's family (e.g. delayed inheritance), they may take longer to achieve a similar socioeconomic status to arranged couples. Reduced financial resources could lead couples to delay reproduction, resulting in longer first birth intervals for eloped couples. Because women and men face asymmetrical financial sanctions for eloping, it is possible that the effect of elopement on reproduction could differ based on whether the brides’, groom's or both families oppose the marriage.

Further investigation of the types and amounts of grandparental investment by marriage type is required to make predictions about fitness effects. Elopements are often accepted following the birth of a grandchild, and this is due to a desire to invest on the part of the grandparents. However, if couples who elope are more likely to live separately from their parents owing to increased tension between daughters- and mothers-in-law or owing to general expectations that eloping couples should be independent, then they may receive less direct grandparental investment simply as a result of distance (rather than the grandparents’ refusal to offer care). Eloped couples may also receive less grandparental investment if they have children later than couples in arranged marriages because the grandparents are older and less vital or even deceased. Perhaps grandparents who cannot invest as much time in directly caring for grandchildren invest in indirect ways, such as financial support for education expenses. Thus, to test whether there are differences in grandparental investment based on spouse choice, dimensions of both direct and indirect investment need to be measured.

D. General discussion

Overall, these discussions indicate a scope for gendered conflict over spouse choice decisions. Cross-culturally, studies demonstrate that women are more likely to have their marriages arranged than men (Broude & Greene, Reference Broude and Greene1983). In this community women appear to have more disagreement with their parents over the qualities of a potential spouse, but at the same time have more socioeconomic benefits to gain from arranged marriage. On the other hand, men in this community appear to have high levels of agreement with their parents on spouse qualities and receive financial and social support regardless of marriage type. For this reason, men's fitness may be less influenced by mode of spouse choice, while women face stronger tradeoffs between the potential genetic fitness benefits of independent mate choice and the fitness benefits derived from the socioeconomic support tied to arranged marriages. If the socioeconomic benefits of arranged marriage outweigh the benefits of independent mate choice, then women entering arranged marriages may be making an optimal choice that is better for their fitness and wellbeing.

The choice to have an arranged marriage, love marriage or elopement may also be heavily dependent on other factors like personality or socioeconomic status. For example, individuals whose personalities are more outgoing or independent may be more likely to choose their own spouses (e.g. Supplementary Information, quote A1), exert more control over fertility decisions, and also argue with in-laws. Additionally, individuals with low socioeconomic status may have less to lose by choosing their own spouse, and thus may be more likely to take that route. If this is the case and marriage types are not randomly distributed (which is likely), then the qualities of the individuals choosing particular types of marriages may be the reason for any fitness differentials, and this may confound the relationships between the genetic, social and financial benefits of different modes of spouse choice and fitness outcomes, as seen in other studies investigating to connections between marriage and fitness outcomes (e.g. Lawson & Gibson, Reference Lawson and Gibson2018). Thus, further research should also investigate the characteristics of the individuals entering into different types of marriages.

E. Limitations

While a mixture of convenience and snowball sampling allowed us to improve group dynamics and increase discussion, it could also limit the diversity of viewpoints. There were also differences in openness across groups. Unmarried individuals, especially women, were more reserved in sharing their opinions about marriage and relationships, possibly because dating and pre-marital relationships are not readily accepted in society. We attempted to circumvent this issue by asking questions about the community and age group generally, but there may still have been some hesitancy to speak openly. Recruitment may have also been more effective for certain socioeconomic or demographic groups. Attempts were made to recruit in various neighbourhoods and at varying times of day to mitigate this, but groups with different demographics may have offered alternative views.

Because many of the questions focused on what is typical in the community rather than personal experiences, we probably missed some variation. While many participants discussed their own experiences unprompted, it is likely that these personal anecdotes reinforced culturally acceptable opinions, while those that do not conform were probably less likely to be shared. We attempted to address this by asking specific questions about socially taboo topics, such as dowry and inter-caste marriages. Likewise, opinions or norms stated in these focus groups may describe ideal situations and not reflect actual behaviour in the community.

Because the ways in which these behaviours can affect reproductive and health outcomes are not always consciously understood, these groups do not offer a direct measure of how marriage type and fitness are connected. Quantitative measures of a wide range of variables would be required to draw conclusions about relationships bwteeen marriage type and fitness. These focus group discussions establish the ethnographic context of marriage in this community. From these data, we can make informed predictions about the potential for and avenues through which spouse choice may influence fitness.