been reported by the nonresponding operations. This 
increased the estimates of totals obtained by the 
respondents and reduced this bias. 
Conceptually, each operation on the sample begins 
the weighting process with an initial weight equal to 
the inverse of the record’s probability of selection. 
Records with sales of $10,000 or more will have an 
initial weight of 1 because they are selected with 
certainty. Records with sales less than $10,000 will 
have an initial weight of about 8. 
If each operation selected for the census provided the 
requested data, the data could simply be multiplied 
by each record’s initial weight then added up to 
attain an estimate for the total amount of the item of 
interest. In the presence of nonresponse, nonresponse 
adjustments are computed and applied to the initial 
weights of the responding operations resulting in a 
nonresponse-adjusted weight greater than the initial 
weight for these operations. The initial weight of 
each nonresponding operation is then adjusted to 
zero. The adjustments are computed in a manner that 
requires the sum of the nonresponse-adjusted 
weights across the responding operations on the 
survey to equal the number of records on the 
sampling frame. 
Nonresponse Weight-Adjustment Groups 
To compute nonresponse adjustments, each 
operation on the mail list was placed in a weight- 
adjustment group. Each operation was assigned to a 
group based on the characteristics used to define the 
group. It was necessary that the characteristics that 
defined the weight- adjustment groups were available 
for responding and nonresponding operations alike. 
Therefore, it was not possible to define weight- 
adjustment groups using data collected via the CHS. 
The information on the sampling frame was used to 
create the weight-adjustment groups and was a 
measure of the horticultural economic size (HES). 
The basic definition of the weight-adjustment groups 
is given below: 
Definition 
HES<$ 10,000 
$ 1 0,000<=HES<$50,000 
$50,000<=HES<$ 1 50,000 
$150,000<=HES<$250,000 
A - 4 Appendix A 
$250,000<=HES<$500,000 
$500,000<=HES 
Must Group (varies by state) 
All records that were considered likely to be very 
large horticultural operations for a given state where 
considered “must” cases and put in a special group. 
For all records in a must group, nonresponse 
adjustment was not allowed and data were imputed 
for any of these records that did not respond. Must 
group definitions varied by state. 
Nonresponse-Adjustment Computation 
A separate nonresponse adjustment was calculated 
within each weight- adjustment group. All 
responding records within each group received the 
same nonresponse adjusted weight. The 
nonresponse- adjustment was obtained by dividing 
the sum of the initial weights across all the records in 
the group by the sum of the initial weights of the 
responding operations in the group. If the sum of the 
initial weights across all records in the group was 50 
and the sum of the initial weights of all responding 
operations in the group was 40, the nonresponse- 
adjustment for the responding operations was 50/40 
or 1.25. The nonresponse-adjusted weight for all 
responding operations in the group was the product 
of the initial weight and the nonresponse adjustment 
of 1.25. This was simply (1 x 1.25). Note that 
1.25*40=50, the sum of the initial weights for all 
records in the group. 
The assumption made when computing nonresponse 
adjustments in this way was that within each weight- 
adjustment group, the data that the nonrespondents 
would have provided had they responded were 
collectively similar to the data provided by the 
respondents. This assumption was made somewhat 
more plausible because operations in the same group 
shared similar characteristics with respect to the 
information used to define the group - the HES. 
Accounting for Misclassification 
When conducting censuses, it is possible that 
respondents might inadvertently report some data in 
error. Operations that really should be determined to 
be in scope for the CHS, i.e., have at least $10,000 
worth of horticultural sales, might report on the CHS 
that their horticultural sales are less than the 
2012 Census of Agriculture 
USDA, National Agricultural Statistics Service 
