To produce usable, representative data from repeated phone surveys on COVID-19 in seven countries across Sub-Saharan Africa, so policymakers can understand the effects on subgroups like women, rural populations and the very poor.
FinMark Trust, a non-profit organization based in South Africa that focuses on financial inclusion in low and middle income countries, asked Mathematica to develop a statistical modelling technique to improve the accuracy of phone survey data so that policymakers can understand the effects of COVID-19 on African populations.
FinMark Trust, with their subcontractor GeoPoll, is undertaking a phone-based survey on the effects of COVID-19 in seven countries across Sub-Saharan Africa: Ghana, Kenya, Nigeria, Rwanda, South Africa, Uganda and Zambia. The survey themes include health and risk behaviors, food security, income, work and job security, personal safety concerns, and access to government and community support.
Mathematica uses an innovative Multi-level regression with post-stratification (MrP) technique to create usable sub-population and nationally-weighted datasets. MrP is a model-based poststratification method that is used to adjust results from a nonrepresentative sample to a target population. It allows to create more representative results and more robust estimations for groups with low mobile phone survey coverage (for example, rural women).
To support dissemination of the survey results and promote data-based approaches to decision making, Mathematica developed a publicly shared blog focusing on the effects of COVID-19 on African populations over time.
Evidence & Insights From This Project
COVID-19 Responses and Impacts: Experiences from Five African Countries
Taking a look at initial survey data from five African countries reveals interesting insights regarding how COVID-19 is shaping access to services and mobility restrictions, coping mechanisms families are using, and early impacts on food security and income.Learn More