Files
Abstract
An area frame sample allocation is developed that will meet the survey cost and precision
requirements of NASS in its estimation of the number of farms not on the mailing
list (NML) for the 2007 Census of Agriculture. Direct and model-predicted estimates of
the stratum standard deviations are obtained for 18 agricultural items (the 8 items used by
NASS for determining the area frame sample and 10 NML items) using the 2002 area frame
sample data, including the supplemental Agricultural Coverage Evaluation Survey (ACES)
segments. Standard deviation estimates of for these 18 agricultural items are considered as
input in the current NASS sample allocation procedure. Three different cases that utilize
these NML items are considered for determining sample allocations. In each case, the previously
available 2007 sample allocation is combined with that obtained for the NML items
to determine a combined area frame sample allocation. The results from each case are compared
to determine the ones that meet the NASS main objectives for the NML farm survey
in 2007. An area frame sample design that includes a supplemental sample of 3091 segments
is recommended.