Each flagged entry was reviewed by staff. Reported 
data that were obviously ineorreet due to 
misinterpretation of a question were either eorreeted 
or deleted prior to the eomputer edit. In some eases, 
respondents may have failed to provide all of the 
information requested, only indieating the presenee 
of an item but not the amount. Some data were 
estimated by the analyst based on other responses in 
the geographie area and by similarly sized farms. 
Prior to publieation, tabulated totals were reviewed 
to identify and resolve remaining ineonsisteneies and 
potential eoverage problems. Comparisons were 
made to 2012 Census of Agrieulture data, 2005 
Census of Aquaeulture data, and other available 
eheek data. The data were proeessed through a 
diselosure program to prevent data from being 
published that eould be soureed baek to an individual 
operation. 
ESTIMATION 
Estimates were produeed for the Nation and for each 
of the 50 States. All respondents to the 2012 Census 
of Agriculture that reported involvement with an 
aquaeulture enterprise, regardless of its eeonomie 
size, were ineluded on the 2013 Census of 
Aquaeulture mailing list. 
The estimation methodology eonsisted of two 
weighting components. The first component was the 
fully adjusted weight pulled in from the 2012 Census 
of Agrieulture. In proeessing the 2012 Census of 
Agrieulture data, statistieal weights were applied to 
eaeh responding reeord. These weights were 
designed to aecount for 2012 Census of Agrieulture 
mail list non-respondents, farms that existed but 
were not ineluded on the 2012 Census of Agrieulture 
mail list, and various farm elassifieation errors. 
The seeond weighting eomponent was from a 2013 
Census of Aquaeulture nonresponse adjustment 
faetor. In spite of a determined effort to obtain 
aquaeulture information from every operation on the 
2013 Census of Aquaeulture mailing list, not all 
operations responded. A nonresponse adjustment 
faetor was used to aecount for active aquaculture 
operations on the list that did not respond to the 2013 
Census of Aquaeulture. 
Together these two weighting eomponents 
A- 2 APPENDIX A 
eompensated for aquaeulture farm data that were not 
obtained from either the 2012 Census of Agrieulture 
or the 2013 Census of Aquaeulture. Eaeh farm on the 
2013 Census of Aquaeulture mail list was put into a 
weight adjustment group. All weight adjustment 
groups were formed within a given State. These 
groups were based on the eeonomie size of the 
farm’s aquaculture enterprise as indicated by the data 
obtained from the 2012 Census of Agrieulture. The 
weights that were earried over from the 2012 Census 
of Agrieulture were summed aeross every reeord 
within eaeh aquaeulture weight adjustment group. 
The resulting weight sum was the best available 
estimate of the number of aquaeulture farms that 
existed for a given State in 2012. The number of 
aquaeulture farms for the weight adjustment group 
was divided equally among all aquaeulture eensus 
respondents within the group. The resulting value 
beeame the statistieally fully adjusted weight for 
eaeh respondent in the weight adjustment group. 
The sum of the adjusted weights aeross all 
respondents in the group neeessarily equaled the 
target value. 
The fully-adjusted weights applied to respondents on 
the 2013 Census of Aquaeulture mail list were 
integerized using a random proeess. This proeess 
rounded eaeh raw weight upwards to the smallest 
integer that exeeeded the fully adjusted raw weight 
using a probability equal to the non-integer portion 
of the raw weight, otherwise, the weight would have 
been rounded downwards to the largest integer that 
was less than the raw weight. 
Example: The raw weight for a reeord is 1.75. It 
will be rounded up to 2.0 with a probability of 0.75 
and rounded down to 1.0 with a probability of 0.25. 
The State total for a partieular eharaeteristie being 
estimated was obtained by multiplying each record’s 
value for the characteristic by the record’s 
integerized weight. The weighted values were then 
summed up over all the responding reeords in that 
State to obtain the State-level estimate. 
RESPONDENT CONFIDENTIALITY 
In keeping with the provisions of Title 7 of the 
United States Code, no data are published that would 
diselose information about the operations of an 
individual farm, unless there is speeifie written 
2012 Census of Agriculture 
USDA, National Agricultural Statistics Service 
