Ghana - Community Services - WASH

The population being studied in this evaluation are the beneficiaries of the water activity. A total of 153,000 people living in 27,407 households in 30 districts of Ghana were affected by the water activity. For the evaluation the target was to conduct household surveys of 600 households in 50 treatment communities and 600 households in 50 control communities.
Out of the 137 program communities, 50 communities were selected at random for the baseline data collection stage. These communities constituted the treatment group. An additional 50 communities not in the program were selected, also at baseline, to serve as the control. These control communities were selected using matching techniques.

At a more granular level, the evaluation includes two data collections: one before the program started (refer as baseline) and a follow-up after the program finished.

During the baseline, in each of the 100 communities, 12 households were selected with the objective of having 10 households in the follow-up study. In each community, a starting point was selected by the supervisor or field manager. This could be a structure like a church, mosque, school, hospital, gas station, police station, etc.; it could also be a bus or taxi park. In any case it was always easy to find or identify. The starting point was used to determine from where the interviewing will begin. The field manager geo-located this starting point and entered the correct information on their community tracking sheet.
Movement from the starting point:
1. Once at the given starting point, interviewers stood with their backs to the structure and moved to the right. Using the day of the interview's date code, and counting households (excluding the starting point), interviewers attempted a contact at the first house corresponding to the day of interview's date code i.e. the date that the interviews were being conducted determined which house was contacted first after the starting point.
2. This household was considered as the first household, and subsequent households were selected by observing the pre-determined skip.3 It is important to keep in mind that we counted individual households and not houses, as a house/building can contain numerous individual households. However, because this study took place in rural areas, a house/building could constitute a household.
3. Interviewers kept to the right side of the road, and did not cross the street.
4. If the interview in a household selected was unsuccessful, interviewers contacted the immediate next household in his/her journey plan; filled in the reason for not being successful in the earlier house and repeated the above process till s/he was successful. Before moving on to a new household the interviewer completed a household identification sheet.
5. When an interview was completed (the interview was successful), then the interviewer observed the skip in his/her journey plan to locate the next household.


In the follow-up survey, the same dwellings that were included in the baseline study were re-visited . We used the complete list of households with contact information and the geographical position of the households from Baseline.

This information help us at least locate the dwelling where the interviewed was done. If the family moved outside of the village, we will consider it a case of attrition. If the family can be found in the same village, we will try to find them.