ESTIMATING CLASSIFICATION ERRORS IN THE AGRICULTURAL CENSUS USING AN AREA FRAME APPROACH

Purpose: Evaluate the feasibility of using NASS' area frame surveys to estimate classification errors in the Agricultural Census.

The Ag Census coverage program currently uses two surveys to measure the accuracy and completeness of farm counts. The NASS area frame surveys are used to measure incompleteness in the Census Mail List, while the Classification Error Survey (CES) uses a small subsample of the mail list to measure duplication and misclassification within the mail list. This research project calls for the use of the 1997 June Area Survey and the 1997 Fall Area Survey to generate State and regional coverage estimates for 11 Western states.

Three questions were added to the June and Fall Area survey questionnaires. These asked for the names and addresses of landlord(s) in the tract, acreage involved in the rental agreement and other addresses and names the tract could have used in 1997. Whole farm information was collected for all tracts. The additional names and addresses were keyed and matched to the Census mail list. Their coverage status was resolved using the completed Census form, the area survey questionnaire and screening form, census research screens, the N on of this approach compared to current census of agriculture procedures and recommendations for possible implementation for the 2002 Census of Agriculture will be made. Final results will be available in the fall of 1999.