Small Area Estimation

NORC statisticians have extensive experience and expertise in using small area estimation methods and models to improve the accuracy of substantive survey estimates. Sample surveys are designed to produce inferential statistics for pre-specified geographic areas or domains. However, often there is interest in developing inferential statistics for smaller geographic areas or finer subpopulations than originally planned at the time of the survey design. NORC researchers have extensive experience using administrative data and other relevant data sources to combine information from multiple sources to improve the accuracy and reliability of small area estimates for these smaller geographic areas and subpopulations.

NORC’s expertise in small area estimation includes:

  • Generating small area estimates for unplanned smaller geographic areas or subpopulations.
  • Generating more reliable small area estimates for the originally planned geographies or subpopulations.
  • Developing sample survey designs which “optimize” generating small area estimates for the geographic areas of interest.
  • Using small area estimation methods to combine estimates from both probability and non-probability samples.
  • Using small area models that capture trends across time and across space.
  • Using area-level and unit-level small area models.