Sample size: planned number of observations
On average each client group has about 25 clients, so the eligible population has about (720 client groups) x (25 clients) = 18,000 clients. We will obtain administrative data on all of these clients, and mini-survey data on all who agree to respond.
We plan to survey 2000 of these ASA clients, through a phone survey:
-1000 loan takers across treatment arms, where loan takers are successively surveyed as they take up the loans, at least 3 months after they first take up the loan. Once any treatment arm reaches 250 surveyed, we will stop surveying loan takers from that arm.
-500 randomly-selected non-loan takers across treatment groups, evenly selected at 125 per group. These will be surveyed so the survey timing balances with the loan taker group. Since a non-loan taker could later become a loan taker, we will randomly replace any respondents who later attrit from this group by taking up a loan.
-500 randomly selected clients in the pure control group. These will be surveyed so the survey timing balances with the loan taker group.