Adolescent Girls Initiative - Kenya (AGI-K)

The goal of AGI-K is to test multi-sectoral packages of interventions in four sectors: violence prevention, education, health and wealth creation. The violence-prevention intervention is a community-level intervention that uses the community dialogues and contracts strategy based on the UNDP/Concern Worldwide model (1). The education intervention is a conditional cash transfer (CCT) based on 80% attendance over the course of a term. The health intervention will follow the Council’s Safe Spaces model (2) in which girls meet in groups once a week under the guidance of a female mentor from the community. The wealth-creation intervention is composed of financial education, which is delivered during Safe Spaces meetings, as well as savings through individual bank accounts (Nairobi) or home banks (Wajir) and a small savings incentives. The following packages of interventions will be tested:
1) Violence Prevention
2) Violence Prevention + Education
3) Violence Prevention + Education + Health
4) Violence Prevention + Education + Health + Wealth Creation

1. UNDP (2005) Leadership for Results UNDP’s response to HIV/AIDS: Community Capacity Enhancement Strategy Note. New York: UNDP. At; http://www.undp.org/HIV/docs/prog_guides/ldp_strategy_note.pdf

2. Austrian K, Ghati D. Girl centered program design: A toolkit to develop, strengthen & expand adolescent girls programs: Population Council; 2010. At; http://www.popcouncil.org/uploads/pdfs/2010PGY_AdolGirlToolkitComplete.pdf

2015-06-22

2017-07-31

The primary outcome measures for the study, to be assessed after 2 years and then again after 4 years, are increased age at first birth, increased age at first sex and increased age at first marriage.

Secondary outcomes include the following:

1. Violence Prevention - Decreased experience of gender-based violence and improved gender norms related to violence

2. Education - Increased mean grade of schooling and increased rate of primary school completion

3. Health - Increased knowledge on sexual and reproductive health and improved decision-making skills

4. Wealth Creation - Increased knowledge on financial education, increased saving and increased participation in income generating activities

The overall, long-term goal of the interventions is to reduce adolescent fertility by improving the well-being of girls in program sites. AGI-K will test multi-sectoral packages of interventions to answer the following research question:

Which combination(s) of interventions produce(s) the greatest change in the well-being of adolescent girls?

Due to the difference in context, cultures, and cost-structures, this study consists of two sub-studies with independent datasets: the Wajir sub-study and the Urban Nairobi Slums sub-study. The research design for the study is a randomized controlled trial (RCT). The unit of randomization is different for the two sites: cluster randomization in Wajir and individual-level randomization in Nairobi. In Wajir, which is less densely populated, a cluster-level design allows for randomization to different combinations of interventions at the community level, while reaching a similar number of girls. In addition, a nonexperimental external control site was included in the design for the Nairobi sub-study to enable comparison of intervention arms with a similar area that did not receive any of the four program interventions.

The project will be evaluated using data from behavioral surveys conducted at baseline, midterm (after 2 years), and endline (after 4 years); a cost-effectiveness and value-for-money costing study; a qualitative study to provide insights into the results shown in the quantitative data regarding the impact of the interventions; and program-monitoring data to track program attendance and participation.

Nairobi - randomization was done using the Excel random number generator. A unique ID was assigned to each girl and the list was randomly ordered, then then divided into 4 groups based on this order. Community stakeholders randomly selected a card with the study arm for each group.

Wajir - randomization was stratified by district: Wajir West (20 clusters), Wajir East (28 clusters) and Wajir South (32 clusters). Using a public lottery method and stratified by district, randomization was done by having a designated community member from each village randomly select a card indicating the study arm.

Nairobi - the randomization was conducted at the individual level.
Wajir - randomization was conducted at the cluster level, stratified by district.

Yes

Nairobi - no clusters (individual randomization)
Wajir - 80 clusters (school-catchment areas) with 20 clusters per arm

Nairobi - 2,400 girls in study site and 600 girls in external control site at baseline; Wajir - 2,160 girls at baseline

Nairobi - 600 girls per arm at baseline
Wajir - 20 clusters (school-catchment areas) per arm and 27 girls per cluster at baseline

Nairobi: Data from the Nairobi Cross-Sectional Slum Survey were used to obtain baseline estimates for childbearing and education in Nairobi. Assuming that 15.4% of girls in the violence prevention only arm would have given birth by endline: we would detect a statistically significant difference of 6.3 percentage points between the violence prevention only arm and each of the other three arms. Assuming a correlation coefficient of 0.33, we would detect a statistically significant difference of 0.55 grade of schooling between any two arms. Wajir: The estimate of the intra-cluster correlation (ICC) and the baseline estimates for childbearing and education were based on data on the Northeastern Province from the 2008/09 Kenya Demographic and Health Survey. Assuming that 17.6% of girls in the violence prevention only arm would have given birth by endline, we would detect a statistically significant difference of 6.9 percentage points between the violence prevention only arm and each of the other three arms. Assuming a correlation coefficient of 0.26, we would detect a statistically significant difference of 0.49 grade of schooling between any two arms.