Strengthening methodological architecture with multiple frames and data sources

Abstract

The United States Department of Agriculture, National Agriculture Statistics Service (NASS) has long conducted statistical surveys and published official statistics about almost every aspect of agriculture. Survey frames include both a list of farm establishments and an area frame, which are often combined to produce multiple-frame estimates. The survey results have also been compared to administrative records on the disposition of commodities. The disposition of commodities, or other intermittent administrative records, which are often received sometime after the survey results are published, must align with published results. For example, hog and pig inventories are published quarterly, while complete slaughter administrative information that matches the inventories is available six to seven months later. NASS has utilized balance sheets and other analysis tools in an expert review process to align the published results to the administrative information. However, the tools lack the statistical rigor that is required by the U.S. Office of Management and Budget (OMB) statistical standards. They are also deficient in the area of statistical measures of error, and are somewhat dependent upon the analyst. This paper will describe efforts to combine the multiple-frame survey results and other data sources using rigorous, statistically defensible methodologies that strengthen the overall results and meet OMB guidelines.