Springs for this study were selected from the universe of 535 unprotected springs in Busia and Butere districts. Technical staff visited each site to determine which springs were suitable for protection. Springs known to be seasonally dry were eliminated, as were sites with upstream contaminants (e.g., latrines, graves). From the remaining suitable springs, 200 were randomly selected (using a computer random number generator) to receive protection. The water quality improvement construction was done by partner NGO, International Child Support, over a period of four years. Springs completed in round 1 and round 2 were considered treatment springs and the rest comparison. This resulted in 91 treatment and 93 comparison springs in the final sample.
Users at each spring were asked to name others who also used the spring. When a household was named by two different respondents, it was designated as a “spring user”. Seven to eight households were randomly selected from the “spring user” list of each spring to be included in the sample. The baseline sample size was 1500 households.
Data on spring water quality, household water quality and household health were collected 3-4 months after round 1 of spring protection, 1 year after round 2 of spring protection and five months after the second survey.