Do adolescent girls' education and friendships have independent effects on early pregnancy? Results of a mediation analysis from a longitudinal cohort study in Nairobi, Kenya

Background: Few studies have examined whether the effect of education on pregnancy and childbearing is due to the academic skills acquired or the social environment that schooling provides. This paper explores whether adolescent girls' learning skills, school enrollment, grade attainment, and friendships affect risk of pregnancy, and whether friendships mediate the relationship between education and pregnancy.

Methods: We draw on three waves of longitudinal data on adolescent girls aged 11-15 in Kibera, an informal settlement in Nairobi, Kenya between years 2015-2019. We use fixed effects regression models to estimate effects of girls' learning skills, school attendance, grade attainment, and friendships on their probability of experiencing a pregnancy. We conduct mediation analyses to assess whether friendships mediate the relationship between education and pregnancy.

Results: By round one (2015), 0.1 % of girls reported having experienced a pregnancy; by round three (2019), 6.3 %. Even after adjusting for friendships, we find that attending school decreases probability of pregnancy by nine percentage points; an additional year of schooling decreases probability of pregnancy by three percentage points; and a one standard deviation increase in numeracy decreases probability of pregnancy by one percentage point. Having any male friends who do not attend school increases girls' probability of experiencing a pregnancy by four percentage points; this association remains after adjusting for girls' education. However, out-of-school girls are far more likely to report out-of-school male friends. We find no evidence that other types of friendships affect girls' probability of becoming pregnant.

Conclusion: We find significant protective effects of school attendance, higher grade attainment and numeracy skills on girls' pregnancy, and that having close friendships with out-of-school males increases girls' probability of pregnancy. We did not find evidence of meaningful mediation, suggesting that the protective effects of school attendance and learning remain regardless of any risk they may face from their friendships.