Estimating nutrition intervention coverage for small areas using Bayesian spatial modeling

Most national surveys produce estimates at national or first administrative levels (e.g. region, state) and may not provide sufficiently precise estimates for smaller geographic units (e.g. district, local area). This limitation constrains the use of data by governments and partners to guide implementation and progress tracking. This technical note presents a flexible step-wise approach for using Bayesian spatial modeling to generate more reliable district- and local-level estimates of intervention coverage from household surveys. A strong foundation in frequentist and Bayesian statistics is necessary to understand and apply the recommended methods.