Education, HIV, and Early Fertility: Experimental Evidence from Kenya

Lowering the barriers to education by providing free school uniforms lowered girls’ and boys’ dropout rates, reduced teen pregnancy within marriage, and decreased teen marriage rates.


In sub-Saharan Africa, sexually transmitted infections (STIs) and an early age of first pregnancy are  two serious health threats to teenage girls. While unprotected sex is the common risk factor for both, policies aimed at reducing unprotected sex must take into account the dichotomy that STIs are more prevalent in casual relationships whereas teen pregnancies are more common in committed relationships. Early pregnancy can also lead to early marriage and prematurely dropping out of school. Taken together with other factors such as the financial burden of schooling, in the Kenyan context, 30 percent of girls who reach sixth grade do not complete 8th grade compared to 21 percent of boys.  

This study seeks to reduce the incidence of STIs and early pregnancy among teens in Western Kenya as well as decrease school dropout rates through a program that provides either an education subsidy in the form of school uniforms, intensive HIV education, or a combination of the two.  In keeping with Kenyan teen culture where condoms are seldom used despite high levels of sexual activity, Kenya’s national HIV/AIDS education curriculum delivered in this study does not focus on prevention through condom use but instead on abstinence until marriage and fidelity. School uniforms serve as an educational subsidy because although education in Kenya is free, school uniforms cost about 1.6% of per capita GDP and can be a considerable burden. 


Subsidies to education costs in the form of free uniforms led to a significant reduction in dropout rates and teen pregnancy within marriage. Abstinence-only HIV education had little effect on STI rates, but led to a decline in teen un-wed teen pregnancy rates. Combining the two programs did not lead to a decline in school dropout rates, teen pregnancy or marriage, though it did reduce girls’ new STI rates.

  • In the standalone education subsidy arm of the study, those in the program were 3.1 percentage points less likely to drop out of primary school than the control group after three years, which represents a 16.5% decrease. Boys in the intervention were 2.5 percentage points less likely to drop out compared to the control group after three years, corresponding to a 19% reduction compared to the control group.
  • Girls in the standalone education subsidy group were 2.7 percentage points less likely to have ever been pregnant in the three years after the intervention than the control group, which is a 17% reduction. After five years, meaning even after the subsidy ended, girls in the intervention group were 4.4 percentage points less likely to have ever been pregnant. The pregnancies that decreased were exclusively within-marriage pregnancy; therefore the intervention did not decrease unwed pregnancy.
  • The education subsidy program alone did not decrease HSV2 (genital herpes) infection.
  • When implemented alone, the HIV education program did not significantly reduce overall teenage pregnancy, HSV infection, and educational attainment for boys or girls. However, unwed pregnancies decreased by 1.4 percentage points (a 30% reduction over the first 3 years) for girls exposed to the HIV education curriculum. This program had no impact on the boys in the intervention group.
  • When the education subsidy and HIV education program are implemented together, there is no significant effect on school dropouts, teen pregnancy or marriage. Girls in the treatment group are 2.3 percentage points less likely to be infected with HSV2 compared to the control group after 7 years, which is a 19% decline.

These mixed results support a model in which the interventions affect the decision to be in committed vs. causal relationships, which affects their final impact. Overall, the results support the idea that fertility and schooling decisions are often made jointly. Lowering obstacles to attending school, like providing school uniforms, has a significant effect on lowering dropout rate and decreasing teen pregnancy within marriage. On the other hand, HIV education that focuses on abstinence until marriage increased early marriage and undermined the reduction in fertility caused by increased access to schooling.


In conjunction with ICS Africa, the Kenyan Ministry of education, the Kenya Institute of Education, and the Kenya National Aide Control Council, two programs aimed at reducing STI, teen pregnancy, and school dropout rates were instituted in Kenya.  The first of these programs provided free school uniforms contingent upon attendance to alleviate educational costs, and the second gave upper primary school instructors training on teaching HIV/AIDS prevention, as set out by national guidelines.  The study followed the implementation and efficacy of these programs in 328 public primary schools throughout seven counties in the Western Kenyan districts of Butero-Mumias and Bungona.  Schools were randomly assigned to one of four groups, resulting in: 82 control schools; 83 uniform subsidy schools; 83 HIV education program schools; and 80 schools that received a combination of both the uniform subsidy and the HIV education program. A total of 19,289 students who were in sixth grade at study onset- 9,487 of which were female- participated. 

Between 2003 and 2007, researchers made school visits to record data on school attendance, marriage or lack thereof, and pregnancy or lack thereof.  A post-experimental survey was administered in 2009 and 2010 to collect follow-up data.   From 2003 to 2007, school visits were made seven times, always unannounced. Dropout and attendance rates were recorded, as were birthing and marriage data.   To confirm accuracy of responses, home visits were randomly paid to 1,420 of the girls’ homes. From 2009 to 2010, long-term data were collected which were representative of 89% of girls and 93% of boys in the sample.  The effect of the respective programs- alone and together- were estimated using regression analysis.

Related GAP Studies