Rainfed water management study

The USAID Feed the Future Innovation Lab for Irrigation and Mechanization Systems wanted to know what would have to change before supplementary irrigation and soil and water conservation were worth a smallholder’s money. In northern Ghana one rainy season carries the year, it has become shorter and harder to predict, and the question of what farmers do about it is usually answered from the outside — from rainfall series and yield gaps rather than from the people farming.

It also had to be answered three times over. IWMI was running the study across three agro-ecological zones with a different team in each. A dataset that was clean for the Upper West but coded differently from the others would have been useless for the only comparison that mattered.

Warc covered the Upper West leg, in the Sudan Savannah. I supervised the data side of it: adapting IWMI’s draft instruments to the local context, pretesting them in the field, and getting them harmonised with the other two regions so the three datasets could be pooled rather than merely filed together.

The instruments were programmed into KoBo Collect and translated into the languages the enumerators would actually be working in. Four enumerators were trained over two days in Tumu, then sent out for a community entry and pilot round — one survey each — before any real collection started, to find the questions people would not answer as written.

Fieldwork ran across five communities in Sissala East and Sissala West. The target was 130 households, twenty-six per community: five above the minimum the analysis needed, on the assumption that some records would not survive quality checks.

Quality control ran daily rather than at the end. Submissions were checked against the protocol as they arrived and the progress dashboard kept current, so an enumerator drifting from the script was corrected the next morning instead of after the round — which is the only point at which it is still fixable.

Alongside the household survey ran key informant interviews with input dealers, commercial and irrigation farmers, the district agriculture directorate and the environmental protection agency, and focus group discussions with smallholders. Crop classification points were collected by GPS inside a watershed boundary IWMI supplied.

The fieldwork finished. Every household survey, interview and discussion was completed, the qualitative material was transcribed, and the quantitative dataset was cleaned and handed over in STATA format with a codebook. Initial analysis and a report went back to IWMI.

Then the funding stopped. USAID cancelled the programme in February 2025 and the study ended before the three-zone comparison it was built for could be run. The Upper West data is complete, clean and documented; the question it was collected to answer is still open.