Theoretical Framework for Integrating Administrative Data and Survey Microdata for Small Area Statistical Estimation in Nigeria

Nigerian national statistical infrastructure increasingly generates administrative data records from health management information systems, tax authorities, and financial regulators that contain population-scale information not exploited by sample survey-based small area estimation methods, creating a theoretical need to formalise integration of these complementary data sources. This dissertation develops a theoretical framework for integrating administrative data and survey microdata for small area statistical estimation in Nigeria, making original contributions to data integration theory and small area estimation methodology. A unified potential outcomes framework for administrative-survey data integration is developed that accounts for coverage error, measurement error differential between data sources, and privacy-preserving record linkage uncertainty. New theoretical results on the bias-variance trade-off in integrated estimators under imperfect administrative data coverage are derived. Optimal integration weights that minimise integrated mean squared error are derived under parametric and nonparametric population models. Three Nigerian application contexts were developed: LGA-level poverty rate estimation integrating NHIA claims and NLSS survey data, LGA-level fertility rate estimation integrating NPC birth registration and NDHS survey data, and LGA-level employment estimation integrating FIRS tax records and NBS labour force survey data. Integrated estimators achieved mean squared error reductions of 35 to 52 percent compared to survey-only Fay-Herriot models across the three applications. Sensitivity analyses quantified how administrative data quality degradation affects integration benefit. The dissertation provides NBS with a theoretically grounded data integration roadmap for the Nigerian official statistics system. Keywords: administrative data integration, small area estimation, survey statistics, Nigeria official statistics, record linkage