Experimental Design
In the 90 blocks included in our study, 30 blocks are assigned to receive no intervention (pure control), 30 blocks are assigned to receive only the linking intervention, and 30 blocks are assigned to receive both the linking and training interventions.
Within the training and linking blocks, half of the GPs are randomly assigned to be NGO-led GPs, while the other half are assigned to be government-led GPs. Although both the NGO- and government-led GPs will receive trainings designed by PRADAN and will have trainers who are trained by PRADAN, the identity and experience level of the the last-mile trainers will vary. In the NGO-led GPs, the trainings are delivered by PRADAN appointed Community Resource Persons (CRPs), locally based employees, typically women, with experience supporting local self-help groups and providing training on a variety of topics. The government-led GPs will instead have trainings that are delivered by government-selected workers nominated by a local MGNREGS official, often "mahila mates." Mahila mates are women who are informally appointed to support community members in providing access to MGNREGS workfare, and typically they have less experience organising trainings. The goal of these two variations is to test whether mahila mates can deliver trainings at the last mile as effectively as more experienced PRADAN trainers, which would make this intervention more scalable.
Within the 30 training and linking blocks, half of the GPs are randomly assigned to be high-saturation GPs, while the other half of the GPs are low-saturation GPs. In low-saturation GPs, we target training approximately 25% of eligible women in the randomly selected village of the GP. In high-saturation GPs, we plan to train approximately 75% (recognizing that we may not quite achieve this target). Population numbers for saturation targets will be created using lists of women who are receiving payments through Ladli Behena Yojana (LBY), a recently rolled-out government program targeting low-income married women between the ages of 21 and 60, which is a good proxy for the target population of our own training intervention.