Reengineering the Estimation and Review Tools at a Federal Statistical Agency
Joseph L. Parsons and Elvera Gleaton
National Agricultural Statistics Service
United States of Agriculture
The National Agricultural Statistics Service (NASS) surveys farmers and ranchers to estimate crop and livestock production. Surveys are also taken to estimate grains stored, environmental and economic items and a host of other survey topics related to United States agriculture. Different estimation and sampling approaches are used for the various reports and surveys. In some instances the data are tabulated and analyzed for reporting errors then weighted for sampling, coverage and nonresponse then published. For many other survey programs, NASS analyzes data from surveys, censuses, and administrative sources using sample survey estimation, graphics, time series analysis, and balance sheets to amalgamate the data into an official estimate. NASS is reengineering its software system used to analyze survey and statistical data as part of a transformational initiative.
In this paper we will report on the planning, design, and implementation of this reengineering effort. NASS previously used a decentralized legacy system of distributed flat databases, Lotus 123, and MapInfo files as the Agency’s estimation and review system. The metadata was stored in multiple FoxPro tables. NASS is merging the disparate systems into a single integrated system, along with adding additional system functionality. The new integrated system, Review Estimates, and Comments, Approval, and Publication (RECAP), enhances analysis, can be easily modified to accommodate commodity program changes, uses standard metadata, is a thin-client application, and simplifies continuity of operations challenges. The new system has nine modules including: a designer tool, comments, estimation, reviewing, mapping, training, charting, messaging, and access control. The goal of the reengineering effort is to reduce the time and errors associated managing the estimation and review process and allow statisticians to more effectively spend their time on activities that improve the quality of NASS data products.