The DHS Program

Design-based Small Area Estimation: An Application to the DHS Surveys (English) 

Document Type

Working Papers

Publication Topic(s)

Fertility and Fertility Preferences, Geographic Information, Infant and Child Mortality

Country(s)

Rwanda

Language

English

Recommended Citation

Recommended citation: Ren, Ruilin. 2021. Design-based Small Area Estimation: An Application to the DHS Surveys. DHS Working Papers No. 180. Rockville, Maryland, USA: ICF.

Download Citation

RIS format / Text format / Endnote format

Publication Date

September 2021

Publication ID

WP180

Abstract:

This study explores design-based small area estimation methods using Demographic and Health Survey (DHS) data collected by The DHS Program, an international program funded by United States Agency for International Development. The DHS surveys are household-based, two-stage cluster surveys that provide key survey indicators for a country’s first level administrative unit, or region. The DHS Program has received increasing requests from host countries for subregional indicator estimates that can be used for policymaking and development planning. Increasing sample size is usually not feasible for meeting this need. One solution is using small area estimation techniques to produce reliable estimation of subregions. This study explores a method for creating a survey domain that covers a small area by pooling clusters or sample units close to the small area from one single target survey or similar surveys conducted in recent years. Close can mean geographically, in time and space, or in other demographic, social, religious, cultural, or economic measures. A survey domain created in this way is easy to analyze with design-based domain analysis tools such as parameter estimation, variance estimation, and confidence intervals for small areas. This study uses data from the 2010 and 2014-15 Rwanda DHS surveys and the proposed methods to produce district-level total fertility rates and childhood mortality rates, which were not provided in the DHS survey reports due to insufficient sample sizes at the district level. The methods described here can be used to produce estimates at the district or subregional level for other surveys and other indicators.