Smoothing County-Level Sampling Variances to Improve Small Area Models’ Outputs

The use of hierarchical Bayesian small area models, which take survey estimates along with auxiliary data as input to produce official statistics, has increased in recent years. Survey for domains are usually unreliable due sample sizes, and the corresponding sampling variances can also be imprecise unreliable. This affects performance model (i.e., will not an estimate or a low-quality modeled estimate), results reduced number statistics published by government agency. To mitigate variances, these survey-estimated typically against direct wherever relationship between two is present. However, this always case. paper explores different alternatives (beyond some threshold) variances. A approach under area-level set-up distribution-free technique based on bootstrap proposed update data. An application county-level corn yield from County Agricultural Production United States Department Agriculture’s (USDA’s) National Statistics Service (NASS) used illustrate approaches. final model-based domains, produced updated each method, compared original 2016.