The study is a cluster randomized control trial. In 8 treatment sectors:
- 64 villages will be randomly selected for treatment, and within treated villages, a total of 128 VSLAs (2/village) will be selected for treatment. Across all treated villages, 640 self-selected couples (5/VSLA) will be selected for treatment. 320 control couples in these 64 treated villages will be surveyed in order to identify spillovers.
- 64 villages will be randomly selected for being control, in these 64 control villages, 960 couples are surveyed, who applied but were not selected because their village was not selected.
Depending on data availability, the villages are matched into 64 pairs on key socio-demographic characteristics and, within each pair, sites are randomly allocated to treatment or control status. Eligible VSLAs and couples will be assigned treatment or control status through a public lottery after baseline is conducted. Within each treatment sector, villages are eligible if they have a minimum of 3 eligible VSLAs, and VSLAs are eligible if they have at least 10 eligible members. VSLA members are eligible for the intervention and the study if they are married or cohabiting with a partner, at least 18 years old, and an existing member of a VSLA. After villages are selected on a computer using a randomization code, VSLAs will then be selected using a public lottery, and in case of oversubscription, the 5 VSLA-member couple volunteers will also be selected through a public lottery. At the time of recruitment and baseline, the participants do not know their own treatment status, nor that of their village or VSLA.
Baseline and follow-up surveys are conducted with all participating couples, village leaders, VSLA leaders, and several randomly selected community members. Differences in outcomes at endline will be attributed to the intervention.