National population mapping from sparse survey data: A hierarchical Bayesian modeling framework to account for...

Area of Science:

• Demography
• Geospatial Analysis
• Statistical Modeling

Background:

• National population and housing censuses are standard for government services but can become outdated, especially in areas with migration or displacement.
• Some countries lack recent census data, hindering planning and resource allocation.
• Accurate, localized population data is crucial for effective public services and development.

Purpose of the Study:

• To develop and demonstrate a hierarchical Bayesian model for estimating population numbers in small areas.
• To provide high-resolution, gridded population estimates with associated uncertainty.
• To offer a solution for population estimation in regions lacking recent census data.

Main Methods:

• Utilized a hierarchical Bayesian model incorporating enumeration data from sample areas.
• Integrated nationwide data on administrative boundaries, building locations, and settlement types.
• Applied the model to estimate population in 10-m grid cells across Nigeria.

Main Results:

• Achieved an overall error rate of 67 people per hectare (mean absolute residuals) or 43% (scaled residuals) in out-of-sample areas.
• Generated gridded population estimates and areal totals with associated uncertainty.
• Demonstrated increased precision for aggregated population totals in larger areas.

Conclusions:

• The developed Bayesian model offers a significant advancement for high-resolution population estimation.
• Provides reliable population data and uncertainty estimates in the absence of complete and recent census information.
• Supports informed decision-making for public services, development, and health campaigns.