Spatially-Explicit Imputation of Missing Data in Small Area Demographic Estimates
Jack Baker, University of New Mexico
Adelamar Alcantara, University of New Mexico
Xiaomin Ruan, University of New Mexico
Small-area estimates of age-specific fertility rely upon incompletely geocoded microdata subject to largely unignorable missingness. In spite of this challenge, little if any discussion of methods for remediating data inputs prior to making ASFR estimates has been presented in the literature. This paper presents a unified , spatially-explicit approach to correcting small area ASFR estimates. The proposed method is based on remediation of birth event counts in light of zip-code level estimates of geocoding success rates and imputation of mother’s age in missing data based on American Community Survey results. Unremediated and remediated ASFR estimates are used to make April 1, 2010 estimates of birth counts during the previous year for Census tracts in the Albuquerque, NM metro area. These estimates are then compared to observed 0 to 1 age counts in the 2010 Census. Results are discussed in light of the growing need for accurate small area demographic estimates.