Bacha posh in Afghanistan: factors associated with raising a girl as a boy

Abstract This paper examines the factors associated with the cultural phenomenon of bacha posh in Afghanistan (in which girls are dressed and raised as boys), which occurs against a background of rigid gender norms and the male-centric nature of Afghan families. Survey data were collected from 1463 women in two provinces of Afghanistan, Kabul and Nangarhar. The primary outcome is a nominal variable, derived from the question, ‘Do you have any girl in your family who has been raised for any time as a boy?’ Independent variables comprise women’s socio-demographic characteristics, family composition, economic characteristics, patriarchal gender attitudes and perceptions of community patriarchal attitudes. Factors associated with bacha posh include women having fewer sons and more daughters, working in the past three months and having less patriarchal gender attitudes. That bacha posh is often driven by a large number of daughters in the family with a corresponding low number of sons suggests that bacha posh is a response to very contextual features of Afghan life, including the preference for sons. Bacha posh in the family is linked to less patriarchal gender norms and can be a way for girls and women to acquire education, mobility and engagement in income-generating activities.


Introduction
Women's autonomy, agency and space to act in Afghanistan have been and remain continually shaped at the confluence of the overarching gender regime, the economic situation, the role of state and international actors and tensions between modernist and traditionalist social and political forces (Ahmed-Gosh 2003;Kandiyoti 2007). There is also much variation in how gender politics play out within individual families and communities (Echavez 2012;Smith 2009). Broadly, however, Afghanistan remains characterised by a deeply patriarchal society with unequal power relations between men and women and dominant modes of hegemonic masculinity (Connell and Messerschmidt 2005). This is particularly visible through strict cultural codes of gender segregation and men's policing of women's mobility and sexuality (AREU 2013). Cislaghi et al. (2019) note that even in highly patriarchal contexts, women can find ways to navigate and resist oppressive power from within the spaces allowed by patriarchal gender systems. In Afghanistan, while women are undoubtedly resourceful in navigating the complex politics and creating spaces for agency and change within these conditions (Billaud 2015;Wimpelmann 2017), in contexts of poverty and gender inequalities, such opportunities are highly constrained. Afghan women continue to negotiate a range of expectations and limitations including demands for social change, respect for and adherence to tradition and potential backlash and reversal of rights (Brodsky 2011).
Women's mobility, particularly in rural settings, is often highly constrained due to concerns about the policing of women's sexuality and maintaining the honour of the family, leading to purdah or the 'protection' and control of women through seclusion and veiling (Boesen 2004). Under the Mujahideen regime (1992)(1993)(1994)(1995)(1996) and Taliban regime (1996)(1997)(1998)(1999)(2000)(2001), women's mobility was tightly controlled (Ahmed-Gosh 2003), particularly under the latter, with the Taliban government introducing and enforcing a universal regulation that women were not permitted to leave their homes without a mahram (male relative) (Ahmed-Gosh 2003;Wimpelmann 2017). While post-Taliban this has tended to ease off, women, particularly in rural areas, are still often expected to be accompanied by a mahram when in public, partly to avoid verbal and physical intimidation, harassment or violence from unrelated men or other community members (AREU 2013;Ganesh 2013). If women have no boys and/or few male relatives, this limits their mobility outside of the home and community, limiting their economic opportunities as well as their ability to socialise or access public services (Echavez 2012;Ganesh 2013).
According to Sindhu (2018), the male-centric texture of Afghan society is particularly visible when examining the construction of the family, where the birth of boys is highly celebrated, and where women are perceived to be responsible for producing male children. There are a number of reasons that boys are highly valued within the family. A key component of the valuing of male children is the role that boys can play in the household's economic productivity. Due to restrictions on the mobility of women and girls, boys can contribute to income generation more easily than female household members. In Afghanistan's patrilineal kinship structures, boys typically stay within their family units when they get older, while women must join the households of their new husbands, and so inheritance is primarily passed through boys (Boesen 2004;Sawitri 2017). Families with no or few male children are perceived to be weak and are severely stigmatised, with women often being blamed for producing 'effeminate' and thus vulnerable families and perceived to place a burden on the family (Nordberg 2014;Sawitri 2017;Sindhu 2018). Such women are mocked for being dokhtar zai, 'she who only brings daughters', and their husbands are in turn mocked for being mada posht, 'he whose woman will only deliver girls' (Nordberg 2014, 16).
It is against this background of the enforcement of patriarchal gender norms and practices and the male-centric nature of Afghan families that bacha posh occurs. The term bacha posh means 'dressed as a boy' and is a cultural practice found predominantly in Afghanistan whereby a girl is raised as a boy, usually for part of her childhood before puberty. Girls who are dressed and treated as boys enjoy mobility and opportunities, such as education, that girls are generally not afforded. Although community members often know that bacha posh children were born as girls, they acknowledge the child as male and 'play along' with the inversion of gender identity (Khodai 2012). However, this playing along becomes more difficult as girls become older and pass through puberty as their identity as a boy becomes more difficult to sustain, although some bacha posh do remain as boys after puberty (Nordberg 2014).
Although there is widespread anecdotal evidence of bacha posh, very little, and only qualitative, research has been conducted on the phenomenon in Afghanistan, with the only substantive piece published being a book named The Underground Girls of Kabul (Nordberg 2014). This book follows the lives of several families that have practised bacha posh. According to Nordberg, there are various reasons for why some families raise girls as boys in Afghanistan. A key one is to escape the social stigma a family can experience from having no boys and the pressure placed on families to bear at least one boy: as Nordberg (2014, 24) describes, 'Having a made-up son was better than none'. Some families raise girls as boys as it is believed to bring good luck for the future births of boys.
In addition, bacha posh are able to support households through employment and income generation. In contexts of high levels of poverty, and limited economic opportunities for women (Ganesh 2013), bacha posh are able to contribute to the household economy and move around communities establishing the networks necessary to do this (Nordberg 2014). Consequently, in households with few boys and many girls, bacha posh can participate in employment and be a practical solution to household barriers to income generation and livelihoods.
Bacha posh can also enable the greater mobility of female household members. They can play the role of mahram by accompanying their sisters, mothers and other female relatives in public spaces. This is particularly the case in more conservative (often rural) settings, where women's mobility is more highly restricted (Nordberg 2014).
Bacha posh may also be driven by a girl expressing what is perceived to be more 'masculine' qualities as a child and wanting to be raised as a boy independently of any economic or family pressures. Sometimes several of these factors intersect, such that a family may require additional boys and will then select a girl who has more masculine characteristics to be the bacha posh (Nordberg 2014).
When girls raised as boys reach puberty, they are usually 'converted' back into girls. This often poses a dilemma for those girls who had more freedom and mobility during childhood, only to have this freedom restricted when being required to re-adopt a feminine identity and sometimes being prepared for marriage a short time after becoming a girl again. Conversion back to being a girl may be particularly difficult for those bacha posh who identify as male and want to continue living as a boy.
There is very little information on the extent to which bacha posh in Afghanistan represents an actual category of transgender subjectivity (or how this subjectivity intersects with sexuality), rather than just a cultural practice driven by the patriarchal setting. It appears that some girls who are raised as boys continue to identify as male (Nordberg 2014), and some may pursue intimate relationships with girls (Billaud 2015;Strochlic 2018). Nordberg's book reveals both anecdotal and empirical examples of bacha posh girls who continued to live as boys and men post-puberty. In one case, a girl was raised as a boy and continued to identify as a boy into adulthood, taking on the responsibility of caring for her mother instead of marrying. Another well-known example of a bacha posh girl who lived as a boy and continues now to live as an adult man is Bibi Hakmeena, a member of the Provincial Council in Khost, who was raised as a boy since the age of 10 at the time of the Soviet invasion in order to assist in protecting her family due to a shortage of older boys and men (Billaud 2015).
Nordberg notes that although bacha posh is mainly a practical solution to having no or few boys, some parents acknowledge the gender-liberating nature of the practice, whereby girls are allowed to experience freedom and the 'other side' of a patriarchal and gender-segregated society, if only for a limited time. And it therefore may be more common in less patriarchal households, or where a woman has previously been bacha posh herself. However, Franks (2014) questions the liberating nature of bacha posh when freedom relies on the masculinisation of female subjects.
Although Nordberg's (2014) book sheds some light on the reasons for bacha posh and the experiences associated with it, no quantitative research has been conducted on the practice's prevalence in Afghanistan, and no quantitative data have been published on the factors associated with it. This paper contributes to our knowledge by presenting quantitative data on the factors associated with bacha posh in two provinces of Afghanistan. We hypothesised, based on the (limited) previous literature on bacha posh and gender relations in Afghanistan, that bacha posh in a family would: (1) be related to the composition of the family, namely number of girl children/boy children, and marriage structure; (2) occur in families with higher levels of poverty and women's engagement in work; and (3) occur in families with higher levels of gender equity as assessed through education and gender attitudes.

Design
This is a secondary analysis of data collected as the baseline of a randomised control trial (RCT) to evaluate a women's economic empowerment and vocational and life skills training programme in Afghanistan, and evaluate whether the intervention can reduce intimate partner violence, strengthen livelihoods and improve women's mental health (Gibbs et al. 2018).
Women were recruited to participate in the trial in six communities in two provinces of Afghanistan: Kabul and Nangarhar. There are a number of differences and similarities between the two provinces and the communities therein. In Nangarhar, the population is predominantly of Pashtun ethnicity, while in Kabul there is more ethnic diversity, with communities comprising a mixture of Tajik, Pashtun and Hazara ethnicities. Although there are important variations in gender relations across different Pashtun provinces in Afghanistan, women in Pashtun communities tend to experience less mobility and social participation than women in other ethnic groups (Azarbaijani-Moghaddam 2012). Despite ethnic differences, there are similarities. In both provinces the communities are either urban or peri-urban and are similar in socio-economic composition. At the time of data collection, there were no seasonal barriers or any specific violent events in the sampled communities that may have amplified women's lack of mobility.
Recruitment of participants was led by the NGO we were working with, using their standard approach for working in these settings that included mobilisation with community and religious leaders to obtain support and access to communities and distribution of information to communities through community meetings and Friday Mosque prayers. Women who were interested in being involved were then asked to come to a central location on a fixed day (Gibbs et al. 2018). Therefore, the sample was self-selecting.
Data were collected through face-to-face interviews conducted by trained Afghan female enumerators in central locations (women's training centres normally) where auditory privacy could be maintained. We used a standardised questionnaire designed for the study, translated into Dari and Pashto languages, and piloted in two provinces of Afghanistan (Kabul and Parwan) in communities where the intervention had previously been implemented.

Measures
The primary outcome is a nominal variable derived from the question 'Do you have any girl in your family who has been raised for any time as a boy?', with a 'yes' or 'no' response. This was included in the survey as one of several items intended to measure women's experiences with cultural practices in Afghanistan related to gender empowerment or disempowerment (other examples included experiences of honour killing in the family).
At baseline, women were also asked a series of socio-demographic questions including age, ethnicity and education level (none, madrasa, primary or secondary).
Women were asked about their family composition and relationship status. Women who were married were asked about the number of living sons and daughters they had and whether they were their husband's only wife or there were other wives.
Women's socio-economic context was measured through questions related to employment, income generation and savings and food insecurity. Women's employment was measured by asking if they had done anything in the last three months to earn money for themselves or their family. Women were also asked to provide a figure for their past month earnings, past month savings and total savings. Household food insecurity was measured through three variables: (1) in the past 4 weeks, how often was there no food to eat of any kind in your house because of a lack of money; (2) in the past 4 weeks how often did you or any member of your household go to sleep hungry because of lack of food; and (3) in the past 4 weeks how often did you or any of your household go a whole day and night without eating because of lack of food? (Coates, Swindale and Bilinsky 2007). Possible responses for all three variables were 'never', 'rarely', 'sometimes' or 'often', scored on a scale from 1 to 4. Scores for the three items were summed to provide a hunger/food insecurity score, with higher scores indicating more food insecurity.
Two patriarchal gender attitudes scales were developed for use in this study, given that 'standard' gender attitudes scales, such as the gender-equitable men scale (GEMS) (Singh, Berm and Barker 2013), would likely not resonate with Afghan women's lives.
One scale assessed women's own patriarchal gender attitudes, and the other measured women's perceptions of their community's patriarchal gender attitudes. Individual patriarchal gender attitudes were assessed through women's agreement with 11 statements related to women's participation, empowerment and subjection to violence. Questions were not framed in relation to actual behaviours but, rather, personal beliefs about what should occur (e.g. 'I think the husbands in my family should respect the opinion of their wives on matters related to income-generating work', or 'I think that a wife in my family who does things that are wrong should be beaten to correct her behaviour'). Items were measured on a four-point scale valued from 1 to 4 ('strongly disagree', 'disagree', 'agree', 'strongly agree') and, after correctly orienting item values, scores were summed to produce an individual patriarchal gender attitudes score, with higher values indicating more patriarchal attitudes. The same 11 items were also included with adjusted wording to represent women's perceptions of community patriarchal gender attitudes (e.g. 'In this community many people think that husbands should respect the opinion of their wives on matters related to incomegenerating work'). The same type of data processing was conducted to produce a community patriarchal gender attitudes score, with higher values indicating women's perceptions of more patriarchal community member attitudes.
The questionnaire included a series of questions related to women's attitudes towards women's participation in different kinds of activities. This included participation in weddings, neighbourhood events, skills training and income-generating activities (e.g. 'I think women should be able to participate in income-generating activities') and women's perceptions of community attitudes towards women's participation in the same activities ('In this community many people think that women should be able to participate in income-generating activities'). Items were measured on a four-point scale valued from 1 to 4 ('strongly disagree', 'disagree', 'agree', 'strongly agree'), and items were summed to create two variables, individual attitudes towards women's participation and perceptions of community attitudes towards women's participation (for both variables, higher scores indicate more gender-equitable attitudes).
We did not collect data at baseline about the relationship between the woman participating in the study and her relationship with the girl raised as a boy. The data were, however, collected at a subsequent point in the trial study (the midline, 12 months after the baseline), where we returned to the same women who had participated at baseline. In addition to the key nominal variable outlined above, the midline survey asked women what their relationship was with the bacha posh in their family. Women could report multiple bacha posh in their family. Although this paper is primarily based on the baseline survey, the findings from the midline survey do provide some additional data that help to contextualise the baseline results.

Data analysis
The data were analysed with STATA 15, and we conducted the analysis separately for women in Kabul and women in Nangarhar. Variables were first summarised descriptively in bivariate analysis as percentages or means, comparing these secondary variables by whether or not a woman reported bacha posh in the family. The statistical test of difference between categories was a Pearson chi-square test for categorical variables and t-tests for continuous variables. The main outcome used in the analysis is the nominal variable indicating if women have ever had any bacha posh in their family.
In order to determine factors associated with bacha posh, binomial logistic regression models were built with the bacha posh variable as the outcome and independent variables drawn from the baseline dataset. We built separate models for each province because of the large number of bacha posh recorded in Nangarhar compared to Kabul. Independent variables were women's socio-demographic characteristics (education level), family composition characteristics (number of living sons, number of living daughters and whether the woman was the only wife of her husband or one of multiple wives), economic characteristics (past month earnings, past month savings, total savings, past three month employment and food insecurity score) and patriarchal gender attitudes (both women's individual attitudes and perceptions of community attitudes). Variables were entered into the models and sequential backwards elimination was used, removing those with the largest p value first until the final variables were retained at p < or ¼ 0.05. Sequential backwards elimination is recommended as it is less likely to incur type II errors (Bingham and Fry 2010). We controlled for age and community.

Ethics
This study was approved by the Institutional Review Board (IRB) of the Ministry of Public Health, Afghanistan, and the Ethics Committee of the South African Medical Research Council. All participants in the research were provided with a verbal and written description of the study and required to provide informed consent before participating in the research.

Results
In total 1396 women were self-selected, eligible and recruited into the study: 1051 from Kabul province and 345 from Nangarhar province. In the baseline survey, 99 women (7.1%) reported that there was a girl in their family who had been raised as a boy. In the midline survey, we found that the most common person to experience bacha posh was the female respondent herself (59.2% of women who reported bacha posh). However, a large proportion of women who reported bacha posh in the family (28.5%) also stated that a daughter had been raised as a boy. Furthermore, 20.7% of women who had been raised as a boy also reported that their daughter had been raised as a boy for a time. Smaller proportions of women who reported bacha posh in the family referred to other types of family members, including a sibling (14.3%), a cousin (10.2%), an aunt (8.2%), a sister-in-law (8.2%), their mother (2%) or any other type of relative (16.3%).
A significantly higher proportion of women in Nangarhar province reported bacha posh (15.4%) compared to women in Kabul province (4.4%) ( Table 1). Significantly more women with Pashtun ethnicity reported bacha posh in their family than Tajik or Hazara women, with a large proportion of women from 'other' ethnic groups (including minority ethnicities such as Turkmen and Uzbek) also reporting bacha posh in their families. However, the effect for Pashtun women is largely accounted for by the majority of Pashtun women reporting bacha posh living in Nangarhar province (51 out of 53). There was no difference in reported bacha posh by the age of women in the study, with just over a third aged 18-24, a quarter 25-34, and just over a third 35-49. Although a slightly higher proportion of women with madrasa education reported bacha posh than women with no education or primary or secondary education, this finding is not significant. Given the significant difference in prevalence of bacha posh between Kabul and Nangarhar provinces, the findings presented further below are disaggregated by province.
There were no significant differences in reports of bacha posh according to women's marital status (Table 2). In Kabul province, having bacha posh in the family was significantly more likely among women with husbands with other wives than women who are the only wife. Women reporting bacha posh in Kabul had a lower mean number of living sons than those not reporting bacha posh; however, this difference was not significant. Women who reported bacha posh also have a significantly higher number of living daughters compared with women who do not report any bacha posh in the family, and this was found in both Kabul and Nangarhar provinces. Regardless of bacha posh, women in Nangarhar had less earnings and savings (monthly and total) and higher hunger/food insecurity scores than women in Kabul (Table 3). Descriptively, women's economic characteristics were found to intersect minimally with bacha posh. In both provinces, working in the past three months, past month earnings or total savings were not significantly related to bacha posh. Women in Kabul who reported bacha posh in their families had significantly higher past month savings than women in Kabul who did not report bacha posh, but the same effect was not found in Nangarhar. When disaggregated by province, no significant relationships were found between bacha posh and household food insecurity.
In Table 4, when disaggregated by province, in Kabul women reporting bacha posh had more equitable individual gender attitudes than women not reporting bacha posh. There were no significant differences between women reporting bacha posh and those who did not, according to perceptions of community gender attitudes, perceptions of community attitudes to women's participation or individual perceptions of women's participation.
The binomial logistic regression models of factors associated with women having a girl in their family raised as a boy, disaggregated by province, are shown in Table 5 (the comparison category is 'no bacha posh'). In Kabul, having fewer living sons, a greater number of living daughters, working in the past three months and less patriarchal individual gender attitudes were significantly associated with women reporting bacha posh in their family. In Nangarhar, only two variables were significantly Table 3. Women's economic characteristics by whether a girl in their family has been raised as a boy, disaggregated by province.

Discussion
Our analysis shows that amongst our self-selecting sample in Kabul and Nangarhar provinces, bacha posh is relatively common. In Kabul one in 20 women in the sample reported bacha posh having occurred in the family, while in Nangarhar it was up to three in 20 women in the sample. A larger proportion of bacha posh was reported in Nangarhar than in Kabul, and among a larger proportion of Pashtun women than women from other ethnic groups, although there was substantial crossover between province and ethnicity, with almost all Pashtun women reporting bacha posh located in Nangarhar. Despite the differences in reported prevalence of bacha posh in the two provinces, the factors associated with reporting bacha posh in the family were relatively consistent, and associated with the numbers of girls, work and gender attitudes in the final regression models. This suggests that bacha posh is not ethnically drive but, rather, is shaped by the intersection of gender inequalities and poverty (Nordberg 2014;Billaud 2015).  As hypothesised, and consistent with the limited literature on this topic (Nordberg 2014), the number of living children reported by women was associated with a bacha posh in the family. In Kabul and Nangarhar, having more daughters was associated with a greater likelihood of bacha posh in the family, and, in Kabul only, having fewer sons was associated with bacha posh. There are two potential explanations for this from previous literature. First, the lack of a boy child in the context of current gender relations in Afghanistan can be seen by families as a significant lack, and bacha posh provides a strategy to overcome this (Nordberg 2014). Second, it may be a response to women's struggles for mobility. This argument is supported by the finding in the bivariate descriptive analysis that among married women in Kabul bacha posh in the family is more likely to occur among women with co-wives than women who are their husband's only wife. In Afghanistan, as in many other Islamic countries, women are often expected to be accompanied by a mahram when in public. In a context in which there are few boys in the household, and/or the male head of household must take responsibility for chaperoning more than one wife, this may limit women's opportunities for mobility. It is possible that raising a girl as a boy can enable additional mahrams to escort women in public, particularly if women have few sons available to assist with this task. However, the presence of co-wives was not significant in the regression model and was subsequently dropped, with the regression model suggesting that number of children is a more important factor. Nevertheless, bacha posh may be a practical response to navigating the gender terrain in Afghanistan.
There were mixed associations between poverty and bacha posh in our analysis. In the regression models, women in Kabul who had worked in the past three months were more likely to report bacha posh in the family, but the same was not observed for women in Nangarhar. Further, descriptively, women in Kabul who reported bacha posh had higher mean savings. The finding in Kabul that women reporting bacha posh are more likely to have worked in the past three months is contrary to previous evidence that suggests that bacha posh in the family is more common when women are unemployed due to the need to mobilise boys for income-generating activities, particularly when there are few sons (Nordberg 2014). However, it is possible that the women in question are able to participate more in labour markets precisely because they have more 'sons' to enable their mobility outside the household. Descriptively, women reporting bacha posh in Kabul had higher past month savings (despite earning less than women who did not report bacha posh), suggesting that girls raised as boys may be making important economic contributions to household savings, either through their own work, or else through enabling women to work. However, as we did not ask about this directly, we cannot provide a conclusive understanding of this.
The lack of association between women's work and bacha posh in Nangarhar may be because, in Nangarhar, more conservative Pashtun cultural norms strongly restrict women's ability to participate in labour markets, as evidenced in the fact that most women in the sample from Nangarhar had not worked in the past three months. As such, even if bacha posh enables greater mobility (Nordberg 2014), this does not necessarily work to increase women's employment as it is effectively prohibited.
Our analysis supports our third hypothesis, that bacha posh is more likely in more gender-equitable families, confirming Nordberg's (2014) argument. In our sample of women from Kabul and Nangarhar, women with a bacha posh in the family reported less patriarchal individual gender attitudes. Previous research on bacha posh suggests two potential reasons for this. First, Nordberg's (2014) work suggests that individuals who raise girls as boys have less conservative gender norms, apparent from their acknowledgement that bacha posh can be experienced as gender-liberating for girls who are able to experience the privileged side of a patriarchal society. Second, from the midline data we found that many women who were themselves bacha posh also had a child who was or had been bacha posh, suggesting that women use it as a way to create some forms of autonomy for their daughters after having experienced it themselves. There remain concerns in the literature that the need to change back to a girl around puberty may have negative mental health effects for girls/women (Nordberg 2014), and that while bacha posh may be individually liberating in the short term, it reinforces overall patriarchal relationships (Franks 2014).

Limitations
This study has a number of limitations. The baseline survey did not specify what kind of relationship female respondents had with the girls in their family reported to have been raised as a boy. However, the inclusion in this paper of midline data on type of family member experiencing bacha posh has helped to partially overcome this. Another gap in the survey tool was the lack of inclusion of a question related to women's perceptions of their own mobility outside the household, which would assist in identifying a more direct relationship between bacha posh and women's mobility. In addition, the overall prevalence of bacha posh is relatively small and, given the overall sample size, limited the power of the analysis to find significant relationships. This paper is based on a secondary analysis of baseline data for an intervention trial, and as such the prevalence of bacha posh is not generalisable to any other populations, and we cannot be sure the extent to which the associations apply outside of this population. We are unsure whether women with bacha posh in the family would be more or less likely to be able to participate in an intervention and research project. Furthermore, given the one-off nature of contact with individuals to complete questionnaires, we recognise this may not produce entirely honest responses, either through social desirability or unwillingness of women to share information about their family outside of the household. Again, we cannot be sure how this would affect the analysis. Finally, as this is a quantitative analysis, we were not able to explore qualitatively more details about the nature of and reasons for bacha posh in the family that women reported on, and whether this supported or hindered women's agency.

Conclusion
The results presented here make an important contribution to understanding the cultural practice of bacha posh in Afghanistan, a topic that has received very little attention in the literature. The paper provides quantitative evidence to support qualitative and anecdotal evidence that suggests that decisions to raise girls as boys are fairly common in Afghanistan. Qualitative research has highlighted how bacha posh emerges at the intersection of particular gender regimes and economic contexts. Our quantitative findings support these arguments. First, it is evident that bacha posh is often driven by a large number of daughters in the family, with corresponding low number of sons, and that bacha posh is therefore a response to very contextual experiences of Afghan lifeeither to enable mobility or overcome the lack of a boy child in the household. Second, our data also support qualitative evidence that suggests that bacha posh in the family is linked to more equitable gender norms. Third, the practice was associated in Kabul with greater engagement in work, and this may be because a bacha posh enables women's greater mobility.