Files
Abstract
The National Agricultural Statistics Service uses its annual June Area Survey (JAS) as the
vehicle to generate annual estimates of farm numbers. These estimates are compared to ones
obtained from the quinquennial Census of Agriculture (conducted for all years ending in 2 and
7). The annual estimate of the number of farms from the JAS has been declining steadily
between censuses, especially between the 2002 and 2007 Censuses. Furthermore, these have
been considerably lower than farm numbers from the census and the difference cannot simply be
attributed to sampling error. Some insight into this issue was provided when the 2007
Classification Error Survey, a census follow-up qualitative study, revealed that agricultural
operations were being incorrectly classified as non-agricultural during the screening procedures
of the JAS.
Therefore, to determine the extent of misclassification resulting from the operational screening
procedures for the June survey and its immediate impact on number of farms estimation, a post-
June Survey intensive screening called the Farm Numbers Research Project was undertaken in
the fall of 2009. The study focused primarily on the newly rotated-in segments which comprised
20 percent of the 2009 JAS sample. It targeted non-agricultural, refusal and inaccessible tracts in
these newly rotated-in segments. The primary purpose of the survey was to verify farm/nonfarm
status, utilizing a survey instrument that contained questions similar to those on the JAS
screening form. The study introduced the concepts of subsampling and subtracts, which are not
routinely part of the JAS. Also, enumerators were instructed to screen residential areas of onehalf
acre or less per parcel, yet another deviation from standard JAS procedures. The results of
this effort are presented in this report.