In order to investigate the relationship between gender, low learning, and subsequent school dropout, we examine how test scores (on cognitive skills assessments) in one survey round are associated with whether the child is still in school in the following round, three to four years later. The Young Lives surveys follow the same children in Ethiopia, India, Peru, and Vietnam for multiple survey rounds, enabling us to link learning and later dropout. For children in the sample who are in school at age 8, we examine the association between test scores at age 8 and whether they are still in school at age 12. Then again, for the children who were still in school at age 12, we examine the association between test scores at age 12 and whether they are still in school at age 15.¹

First, descriptive statistics show that girls in our sample are less likely to drop out of school than boys (Figure 1). Between ages 8 and 12, 1.3 percent of girls dropped out versus 2.2 percent of boys. Dropout is higher for both genders in the second period: between ages 12 and 15, 7.1 percent of girls dropped out, compared with 8.5 percent of boys. Furthermore, we find no statistically significant differences in boys’ and girls’ performance on cognitive tests; they perform equally well at both ages we analyse.

Table 1 shows gender disaggregated regression results. First, we find a strong, significant relationship between achievement on mathematics tests at age 8 and the odds of dropping out of school four years later for both girls and boys.  For girls, a one standard deviation higher math score at age 8 is associated with a 49 percent reduction in the odds of dropping out by age 12. For boys, one standard deviation higher math score at age 8 is associated with a 56 percent reduction in the odds of dropping out by age 12. We find a similar, strong, and significant relationship between math test scores and dropout between the ages of 12 and 15 – in both genders, a one standard deviation higher math score is associated with a 49 percent reduction in the odds of dropout.

Second, the regressions show that after controlling for other factors in our data, such as rural/urban geography, pre-primary attendance, household wealth, and household head education, girls in this sample are still less likely to drop out of school between ages 8 and 15 than boys. Between ages 8 and 12, being a girl is associated with a 39 percent reduction in the odds of school dropout when we control for these other factors. Between the ages of 12 and 15, being a girl is associated with a 22 percent reduction in the odds of school dropout, though this result has slightly lower statistical significance (p < 0.1).

Taken together, these results show a strong relationship between low learning and school dropout for both genders. Further, they suggest that even after controlling for other commonly cited factors precipitating dropout, girls are still less likely to drop out of school than boys (suggesting efforts to keep girls in school should perhaps expand to also help keep boys in school).

Table 1: Gender Disaggregated Regression results

Qualitative findings suggest that low learning contributes to girls dropping out of school in multiple ways. Rather than passively failing out of school, girls often exercise a great deal of agency in their decision to drop out, utilising their own assessments of their learning and the quality of education provision to make a dropout decision. Biritu in Ethiopia, for example, recognised the shortcomings in the education she was provided, and decided not to continue: “When I evaluated my grades in the past seven grades, I found out that I was not able to sit for the next year national examination because we were not taught very well.”² Fifteen-year-old Esmeralda in Peru felt shame from failing classes and repeating grades, and eventually dropped out: “Having failed two courses that she needed to progress to the third grade, she left the school because she did not want to repeat the grade. She expressed being ashamed of re-doing a grade with younger children, particularly in the same group as her younger brother.”³

In other cases, parents chose to discontinue their daughter’s schooling because of low learning; while parents often blamed what they perceived as their child’s low inherent abilities, researchers place blame on the quality of the education being provided by the education system. In Vietnam, Ho Nit’s parents thought she was slow and struggled to remember what she was taught: “The letters know her but she does not know the letters,” lamented her mother.⁴ Similarly, although Long wanted to continue her studies, when she failed her Grade 9 exams her parents concluded that her innate academic capabilities were incompatible with further schooling.⁵

The qualitative findings also reveal that low learning often underlies other more commonly cited reasons girls drop out of school, such as marriage and work. In Ethiopia, while Beletech’s caretaker wanted her to continue schooling instead of marrying, she decided to elope after losing interest in school following years of low learning.⁶ Qualitative findings across the Young Lives dataset indicate that girls are more likely to elope or have early marriages arranged when low learning has already led them to drop out.

Furthermore, girls commonly feel pressure to secure a future for themselves when they see that they are learning little in school, and they often see marriage as a future-proofing strategy.⁷ When Fatuma, in Ethiopia, learned she had failed the national exam, “she started devising other coping mechanisms for survival … After realizing that neither further education nor jobs were forthcoming, Fatuma began to envisage getting married. She soon married and had her first child the next year.”⁸ Girls similarly see work as a way to provide for their future. In Ethiopia, a sixteen-year-old girl explained that she was learning embroidery and sewing from a local mosque in addition to her formal education, as she was worried that she might fail her national exam and be perceived as a burden to her family.⁹

These findings add to the growing body of evidence showing that girls’ learning is critical to ensuring that they realise the full benefits of education. The much-touted health and social benefits of girls’ education, such as lower child mortality and fertility rates, are in part driven by girls achieving foundational skills like literacy, and as this new RISE paper shows, girls who are learning are more likely to stay in school. The international community has set ambitious and mutually reinforcing goals for girls’ education. Focusing on girls’ learning will be key to making both goals a reality.