The authors have declared that no competing interests exist.
Current address: UNICEF Nepal, Kathmandu, Nepal
School dropout and child marriage are interrelated outcomes that have an enormous impact on adolescent girls. However, the literature reveals gaps in the empirical evidence on the link between child marriage and the dropout of girls from school. This study identifies the ‘tipping point’ school grades in Nepal when the risk of dropout due to marriage is highest, measures the effect of child marriage on girls’ school dropout rates, and assesses associated risk factors. Weighted percentages were calculated to examine the grades at highest risk and the distribution of reasons for discontinuing school. Using the Nepal Multiple Indicator Cluster Survey (MICS) 2014 data, we estimated the effect of marriage on school attendance and dropout among girls aged 15–17 by constructing logistic regression models. A multivariate logistic regression model was used to assess risk factors of school dropout due to child marriage. It was found that early marriage is the most common reason given for leaving school. Overall, the risk of school dropout due to marriage heightens after girls complete the fifth or sixth grade. The risk of girls’ dropping out peaks in the seventh and eighth grades and remains noteworthy in the ninth and tenth grades. Married girls in Nepal are 10 times more likely to drop out than their unmarried peers. Little or no education of the household head, belonging to the Kirat religion, and membership of a traditionally disadvantaged social class each elevate the risk of school dropout due to early marriage. The findings underscore the need to delay girl’s marriage so as to reduce girls’ school dropout in Nepal. School-based programmes aimed at preventing child marriage should target girls from the fifth grade because they are at increased risk of dropping out, as well as prioritizing girls from disadvantaged groups.
School dropout and child marriage are interrelated outcomes that have an enormous impact on adolescent girls, curtailing full realization of their rights, limiting their livelihood options, and harming their health and wellbeing as well as that of their children. There are also broader social implications related to economic development and gender equality. Education is a fundamental human right for all children, as enshrined in the Convention on the Rights of the Child (1989), and is seen as an enabling right essential for the fulfilment of other rights thanks to the empowering impact of education on societies and individuals [
Nearly 58 million children of primary school age (typically between 6 and 11 years) around the world were not enrolled in school in 2012 [
While boys often face economic pressures to drop out of school related to migration and labour, child marriage is globally considered a major barrier to retaining girls in school. South Asia has some of the highest rates of child marriage in the world, with more than one third of all child marriages that occur globally occurring in the region [
It is argued that education is a powerful positive predictor of female age at marriage [
However, association does not equate to causation. Although the strength of association is one of the Bradford Hill criteria used to make an overall judgement about a causal relationship [
Decisions on these life events are complex, multifaceted, and most often private family decisions. As Nguyen [
This paper assesses the effect of child marriage on school dropout among Nepali girls. The four important questions that this study sought to answer are as follows: (i) How common is school dropout due to marriage among girls? (ii) What is the ‘tipping point’ grade at which the risk of leaving school due to marriage escalates? (iii) How much more likely (or less likely) is school attendance and school dropout among married girls compared to their unmarried peers? (iv) Who is more likely to leave school due to marriage?
Data used in this study were extracted from the Nepal Multiple Indicator Cluster Survey (MICS) 2014 [
To address our multiple research questions, we used two subsamples of the MICS data for analyses. First, analysis of the prevalence of school dropout due to child marriage and analysis of the grade of school in which girls are at the highest risk of leaving due to marriage were both limited to women aged 20–24. We chose this age range because women are either no longer enrolled or have completed secondary school by age 20. Second, to examine the effect of child marriage on school attendance and school dropout, as well as to assess risk factors associated with school dropout due to marriage, we restricted analyses to girls aged 15–17. Focus on older adolescents is appropriate, as the large majority of those married as minors in Nepal are in this age range. Furthermore, the MICS data do not include respondents below the age of 15.
This study began by looking at the reason for discontinuing education using descriptive statistics. The reason for school dropout was a categorical variable which was assessed by the question, ‘What was the main reason you did not continue your studies further?’ The possible reasons provided in the questionnaire were ‘economic reasons’, ‘parents did not allow’, ‘got married’, ‘school facility far away’, ‘need to do household work’, ‘did not like to study’, ‘physical disability’, and ‘other’. We also analysed the highest grade completed by women who left school due to child marriage. The highest grade was treated as an interval variable in the analysis.
We examined three dependent variables of interest, using regression analyses. Firstly, the school attendance variable dichotomized whether girls were attending school or not when the survey was conducted. This variable was coded as ‘1’ if a girl was attending school, ‘0’ otherwise. Second, the school dropout variable also dichotomized whether girls withdrew from school before completing secondary education (the twelfth grade, or approximately 17 years old, in the context of Nepal). This variable was coded as ‘1’ if a girl was out of school, ‘0’ otherwise. Grade repetition was not taken into account in the variable. These two variables were assessed using the question, ‘Are you currently enrolled in any school?’
Another dependent variable differentiated between unmarried girls attending school and married girls who left school due to marriage. Married girls who left school and reported marriage as the main reason for leaving school were coded as ‘1’, and girls who were unmarried and attending school at the time of the survey were coded as ‘0’. Girls who left school for other reasons were dropped from this variable.
Child marriage was used as the independent variable in regression analyses on school participation and school dropout. A dichotomous variable was made for the variable based on reported age at first marriage. In this study, a girl was considered married if she was formally married or informally in union before the age of 18.
The covariate variables used in the multivariate logistic regression models were demographic and socioeconomic characteristics which were assessed by questions within the MICS questionnaire. The covariate variables include:
age
place of residence (urban vs. rural)
household wealth status
religion
social class
the level of education of the household head.
A relative index of household wealth, which represents economic status, was calculated based on a standard set of interviewer-observed assets, including the ownership of consumer items and dwelling characteristics. Household wealth status was determined by a wealth score given to each household in the total sample based on the assets owned by that household; women were then divided into poor (lower 40%) and non-poor (upper 60%). Religion was classed as Hindu, Buddhist, Muslim, Kirat, Christian, and others. Social class as defined by caste was dichotomised as Dalit or non-Dalit based on ethnicity data. The education level of the household head was classed according to the highest level of education attended ranging from no education to primary education, and secondary or higher education.
Data analyses in this study comprise descriptive analyses and logistic regression analyses. Weighted percentages were calculated to examine the distribution of the reasons for leaving school, and the highest grade of school dropout due to child marriage. Similarly, weighted percentages were calculated to compare demographic and socioeconomic characteristics among married and unmarried girls. Pearson’s chi-squared test was used for categorical variables to determine whether differences in demographic and socioeconomic characteristics were statistically significant between married and unmarried girls. Student’s t-test was used for continuous variables for the same purpose. We considered a two-tailed p-value of < 0.05 to be statistically significant.
The likelihood of school attendance and school dropout for married girls was assessed by constructing logistic regression models and estimating unadjusted odds ratios for the dependent variables of school participation and school dropout. We regressed each dependent variable against child marriage. Further, adjusted odds ratios were estimated after controlling for demographic and socioeconomic characteristics (age, place of residence, household wealth status, religion, social class, and education of the household head). Prior to calculating adjusted odds ratios, Pearson's correlation coefficients were run to test possible collinearity among these covariates.
By constructing a multivariate logistic regression model, the likelihood of unmarried girls attending school or married girls having left school due to child marriage was assessed to identify the predictor of school dropout due to child marriage. We regressed the girl’s status against age, place of residence, household wealth status, religion, social class, and education of the household head, which were treated as independent variables in this model.
All the data were weighted to account for the complex sample design, such as stratified sampling and probabilities of unequal sample selection between regions. Using the national women’s weighting for the entire MICS allows for analyses that produce nationally representative estimates. All statistical analyses for this study were conducted using Stata 13 (StataCorp LP, College Station, TX).
The Nepal MICS procedures were reviewed and approved by the Nepal Central Bureau of Statistics and UNICEF. Because this study involves secondary data analyses of a publicly available dataset which does not reveal the identity of the respondents, ethical approval from respective institutions was not required.
Reason | % |
---|---|
Got married | 39.8 |
Did not like to study | 21.7 |
Economic reasons | 15.3 |
Need to do household work | 14.9 |
Parents did not allow | 5.7 |
School facility far away | 2.0 |
Physical disability | 0.7 |
Data are given in weighted percentages.
Grade | % |
---|---|
1 | 0.0 |
2 | 0.3 |
3 | 1.6 |
4 | 5.1 |
5 | 11.4 |
6 | 11.1 |
7 | 21.0 |
8 | 21.8 |
9 | 13.8 |
10 | 13.8 |
Data are given in weighted percentages.
The Nepal MICS data generated a subsample of 1,631 girls aged 15–17. Participants reported a mean age of 16.0 years (SD = 0.79). Descriptive statistics for the demographic and socioeconomic characteristics of the girls included in this study are presented in
Participants (n = 1,631) | Married (n = 202) | Unmarried (n = 1,429) | P-value for difference in covariates by marital status | |||||
---|---|---|---|---|---|---|---|---|
Married | ||||||||
Yes | 202 | (11.8%) | - | - | - | - | ||
No | 1,429 | (88.2%) | - | - | - | - | ||
Schooling |
P < 0.001 | |||||||
In-school | 1,093 | (78.5%) | 53 | (31.2%) | 1,040 | (84.9%) | ||
Out-of-school | 258 | (21.5%) | 111 | (68.8%) | 147 | (15.1%) | ||
Highest level of education | P < 0.001 | |||||||
None | 73 | (5.8%) | 23 | (15.0%) | 50 | (4.6%) | ||
Primary | 180 | (11.0%) | 38 | (20.2%) | 142 | (9.8%) | ||
Secondary | 1,178 | (69.2%) | 127 | (60.4%) | 1,051 | (70.4%) | ||
Higher | 200 | (14.0%) | 14 | (4.3%) | 186 | (15.2%) | ||
Current age | P < 0.001 | |||||||
15 | 520 | (30.9%) | 22 | (13.9%) | 498 | (33.2%) | ||
16 | 613 | (39.0%) | 68 | (32.9%) | 545 | (39.8%) | ||
17 | 498 | (30.2%) | 112 | (53.2%) | 386 | (27.1%) | ||
Place of residence | P = 0.004 | |||||||
Urban | 307 | (14.2%) | 30 | (7.7%) | 277 | (15.1%) | ||
Rural | 1,324 | (85.8%) | 172 | (92.3%) | 1,152 | (84.9%) | ||
Ecological zone | P = 0.806 | |||||||
Mountain | 449 | (7.0%) | 53 | (6.2%) | 396 | (7.2%) | ||
Hill | 606 | (42.6%) | 84 | (43.9%) | 522 | (42.5%) | ||
Terai | 576 | (50.3%) | 65 | (50.0%) | 511 | (50.4%) | ||
Household wealth status | P = 0.072 | |||||||
Poor | 885 | (42.7%) | 127 | (50.1%) | 758 | (41.7%) | ||
Non-poor | 746 | (57.3%) | 75 | (49.9%) | 671 | (58.3%) | ||
Religion | P = 0.058 | |||||||
Hindu | 1,332 | (83.7%) | 164 | (81.1%) | 1,168 | (84.0%) | ||
Buddhist | 155 | (7.00%) | 12 | (4.7%) | 143 | (7.3%) | ||
Muslim | 41 | (4.1%) | 8 | (7.9%) | 33 | (3.6%) | ||
Kirat | 65 | (2.8%) | 16 | (4.9%) | 49 | (2.5%) | ||
Christian | 34 | (2.3%) | 2 | (1.4%) | 32 | (2.4%) | ||
Others | 4 | (0.2%) | 0 | (0%) | 4 | (0.2%) | ||
Social class |
P < 0.001 | |||||||
Non-Dalit | 1,437 | (88.5%) | 157 | (76.9%) | 1,280 | (90.1%) | ||
Dalit | 192 | (11.5%) | 45 | (23.1%) | 147 | (10.0%) | ||
Education of household head |
P < 0.001 | |||||||
None | 738 | (45.1%) | 113 | (60.3%) | 625 | (43.0%) | ||
Primary | 341 | (21.5%) | 51 | (24.5%) | 290 | (21.1%) | ||
Secondary or higher | 545 | (33.3%) | 38 | (15.2%) | 507 | (35.8%) |
Data are n (weighted %).
Absolute number of participants does not perfectly correspond to percentages presented because weighted analyses were used.
a n = 1,351
b n = 1,629
c n = 1,624
There were significant differences in the demographic and socioeconomic characteristics between girls aged 15–17 who were married and unmarried. Women who were married at a young age were more likely to have dropped out of school, have no formal education, reside in a rural setting, live in poverty, or have an uneducated household head, and they tended to be older adolescents, Dalit, or Muslim.
Married | Unmarried | |||
---|---|---|---|---|
In school | 53 | (3.9) | 1,040 | (77.0) |
Out of school | 111 | (8.2) | 147 | (10.9) |
Data are n (weighted %).
OR | 95% CI | Adjusted OR |
95% CI | |
---|---|---|---|---|
Married | 0.08 |
[0.05, 0.13] | 0.10 |
[0.06, 0.17] |
n = 1351 for OR; n = 1344 for adjusted OR. OR = odds ratio; CI = confidence interval
a Analysis adjusted for age, place of residence, household wealth status, religion, social class, and education of the household head.
*** Level of significance
The logistic regression analysis also shows that married girls were significantly more likely to leave school (OR 12.40; 95% CI 7.54–20.40) compared to their unmarried peers (
Unadjusted OR | 95% CI | Adjusted OR |
95% CI | |
---|---|---|---|---|
Married | 12.40 |
[7.54, 20.40] | 10.04 |
[5.84, 17.25] |
n = 1351 for OR; n = 1344 for adjusted OR. OR = odds ratio; CI = confidence interval
a Analysis adjusted for age, place of residence, household wealth status, religion, social class, and education of the household head.
*** Level of significance
Multivariate logistic regression analysis for the status of girls aged 15–17 (unmarried and attending school vs. having left school due to marriage) reveals that the age of girls, the level of education of the household head, and religion were the main contributing factors to school dropout due to child marriage (
Variable | Adjusted ORs | 95% CI | |
---|---|---|---|
Age |
2.81 |
[1.73, 4.56] | |
Place of residence |
|||
Rural | Reference | ||
Urban | 1.18 | [0.48, 2.92] | |
Household wealth status |
|||
Non-poor | Reference | ||
Poor | 0.95 | [0.52, 1.76] | |
Religion |
|||
Hindu | Reference | ||
Buddhist | 0.77 | [0.14, 4.39] | |
Muslim | - | - | |
Kirat | 2.31 |
[1.00, 5.34] | |
Christian | 0.46 | [0.07, 2.97] | |
Others | - | - | |
Social class |
|||
Non-Dalit | Reference | ||
Dalit | 2.24 | [0.92, 5.43] | |
Education of the household head |
|||
Secondary or higher | Reference | ||
Primary | 4.56 |
[1.65, 12.61] | |
No education | 4.44 |
[1.57, 12.56] |
Note. OR = odds ratio; CI = confidence interval
a Analysis adjusted for place of residence, household wealth status, religion, social class, and education of the household head
b Analysis adjusted for age, household wealth status, religion, social class, and education of the household head
c Analysis adjusted for age, place of residence, religion, social class, and education of the household head
d Analysis adjusted for age, place of residence, household wealth status, social class, and education of the household head
e Analysis adjusted for age, place of residence, household wealth status, religion, and education of the household head
f Analysis adjusted for age, place of residence, household wealth status, religion, and social class
* Level of significance
The current study shows that marriage was the most common reason given for girls’ school dropout. Overall, the risk of school dropout due to marriage heightens after girls complete the fifth or sixth grade. The risk of girls dropping out peaks in the seventh and eighth grades and remains high in the ninth and tenth grades. The analysis also demonstrates that married girls aged 15–17 in Nepal are 10 times more likely to leave school than are their unmarried peers. Little or no education of the household head, belonging to the Kirat religion or historically disadvantaged social class elevate the risk of school dropout due to child marriage. Girls living in a household with a household head who had not attended school or who had only attended primary school are approximately 4.5 times more likely to leave school due to child marriage, compared to having a household head who attended secondary school or higher. Girls belonging to the Kirat religion are 2.3 times more likely to drop out of school due to child marriage than their Hindu peers. Dalit girls are 2.2 times more likely to leave school due to child marriage, compared to their peers belonging to other castes.
The very high percentage (40%) of Nepali women reporting marriage as the main reason for school dropout is exceptional. The percentage of women aged 20–24 in sub-Saharan African countries who reported early marriage as the main reason ranged from 2% in Togo, Benin, and Côte d'Ivoire to 28% in Chad [
These findings on the significant effect of child marriage on girls’ education is consistent with those of earlier studies [
The risk of school dropout due to early marriage is exacerbated among girls living in a household with a household head with little or no formal education, from historically disadvantaged social class, and belonging to a certain religion. The finding that the level of education of the household head has a significant influence on the probability of school dropout due to child marriage demonstrates the inter-generational impact of educational attainment. Furthermore, as close to 30% of household heads in Nepal are women [
Evidence generated from this study can aid policy makers and education planners by helping them better understand the magnitude of the effect of child marriage on school attendance, who is at the highest risk of school dropout due to child marriage, and in which grades they face the highest risk of dropping out. The evidence could potentially inform policy making and advocacy for eliminating child marriage. Nepal has pledged to end child marriage by 2030 [
The present study has several limitations that should be considered. First, the MICS data are prone to recall and social desirability biases because of their use of self-reported data. Possibly, the illegality of child marriage may have led to systematic under-reporting because of fear of legal repercussions. If underreporting of child marriage was the case, the study may have underestimated the effect of child marriage. In Bangladesh, however, a study found widespread misreporting of age of marriage, with more than half of women reporting their age at first marriage as younger than it actually was [
Second, the cross-sectional nature of the MICS data does not permit the investigation of sequential events and causality. Age at the time of leaving school was not gathered in the Nepal MICS, so it was not possible to establish whether or not school leaving preceded marriage. Given the strength of the association and the frequency of reporting marriage as the main reason for school dropout, however, it is reasonable to argue that child marriage is a key factor that leads girls to leave school. Third, the household wealth index was based on assets of the current household that girls live in. For married girls, household wealth assessed in the survey did not consider the parental household. The potential inconsistency of economic status may have introduced an unsystematic bias in the analyses. Similarly, the education data of the household head was based on the household that married girls were living in at the time of the survey. Fourth, the analyses of the risk of school attendance and school dropout were limited to girls aged 15–17, and therefore the findings are not generalizable to other age groups in the country.
The strength of this study is that it deals with nationally representative data which was generated by a well-established national survey that used standardized measures and protocols. On the other hand, using panel data that track girls over time is a more credible way of establishing sequential events and causality between school dropout and child marriage. While not addressed in this study, further work to understand the impact of child marriage on boys who marry early is also needed.
The current study yields solid evidence that child marriage significantly increases girls’ risk of school dropout in Nepal. Despite the limitations discussed above, the strength of the association and the frequency of reporting of marriage as the main reason for school dropout are remarkable enough to warrant the conclusion that child marriage is a main driver of girls’ dropping out of school in Nepal. This finding underscores the need to delay marriage in order to reduce gender disparity against females in school retention and the out-of-school population in many parts of the country. Reducing the risk of dropout requires strategies that retain girls in school and facilitate a smooth transition to secondary education. School-based programmes aimed at preventing child marriage should target girls from the fifth grade because of their escalated risk, and they need to prioritize girls from disadvantaged groups.
(DOCX)
(DOCX)
The authors would like to thank Ms Kendra Gregson, Mr Ivan Coursac and Ms Rhiannon James, of the UNICEF Regional Office for South Asia, and Ms Mariana Muzzi of UNICEF Nepal, for their valuable feedback.