Agricultural Survey for Cocoa Sector Evaluation in Liberia

IFAD’s Tree Crops Extension Project (TCEP) invested in revitalizing cocoa plantations and improving smallholder livelihoods in Liberia’s Nimba County. As part of IFAD’s evaluation cycle, the Evidence for Development Impact (EDI) office needed reliable survey data to assess the project’s outcomes, and needed it fast.

rowsquared was selected to deliver this on a tight timeline, working together with local partner Q&A. Our task was to design and implement a household survey that could capture the complexities of smallholder cocoa farming in rural Liberia.

A central challenge was measuring cocoa production accurately. Farmers harvest, ferment, dry, and sell cocoa in multiple batches, through different channels, and with varying levels of processing. Standard survey modules do not capture this well. To address this, rowsquared developed a custom-made cocoa module that traces the full production cycle, from harvest quantities and fermentation practices to drying methods, individual sales transactions, and buyer relationships.

We adapted IFAD’s standard survey instruments to the local context and the specific interventions of TCEP, including questions on plantation rehabilitation, nursery participation, cooperative membership, and access to project-provided inputs such as seedlings.

In a dedicated pre-test, we verified that the new cocoa module and the adapted instruments were fully functional in the field. This step was critical: reporting patterns for harvest quantities, processing volumes, and sales were not straightforward, and the pre-test allowed us to refine the tools so that enumerators and respondents could navigate them reliably.

We then conducted a comprehensive interviewer training, including practical field exercises, to prepare the team for the demanding fieldwork conditions in Nimba County, where internet connectivity was limited and logistics were challenging.

rowsquared managed the full fieldwork operation and ran a rigorous quality assurance process throughout, including high-frequency data checks and close monitoring of data consistency. Despite the tight timeline, we delivered a clean, high-quality dataset on schedule.