Malawi’s last population and housing census was conducted in 2018. The National Statistical Office (NSO) reported that approximately 95% of households were georeferenced, and digital enumeration areas (EAs) had been developed for the entire country. As Malawi plans for the next census, expected in 2028, the NSO has indicated that due to limited funding, a full national household listing before census will not be feasible. Instead, they are exploring innovative approaches to produce population estimates that closely approximate ground data and can be used to update existing census EAs and support census planning.

The NSO indicated that for the 2018 census, EA demarcation was conducted manually. It was therefore planned that they would manually update their EAs for the 2028 census. This manual approach has been recognized as both time-consuming and resource intensive. As a result, the Malawi NSO is eager to explore automated methods that could enhance efficiency and reduce costs.

To address these challenges, the UK Office for National Statistics (ONS) has coordinated a series of discussions and technical meetings with WorldPop and the Malawi NSO. The aim has been to introduce and explore geospatial tools and modelling techniques that can support census preparation. With our distinguished expertise in geospatial modelling, WorldPop will lead the integration of recent survey data with the 2018 census data into a machine learning-based population model. This model will incorporate a wide range of geospatial layers derived from satellite imagery, including the latest Google building footprints, to estimate the number of households across Malawi.

The initial model will serve as a key input for updating a pilot set of census Enumeration Areas (EAs) using the preEA tool. This pilot will then be validated on the ground by Malawi’s National Statistical Office. The data collected during this validation phase will be used to refine and finalize the model in preparation for operational implementation and planning for the next census.

This process aims to serve as a proof of concept, demonstrating a cost-effective, data-driven approach that can be replicated in other countries—especially as we approach the 2030 census round, where many nations face financial constraints.

Link image

Lilongwe, Malawi by N J Spicer, 2006 CC BY 2.0