were ineluded in the 2012 CML. A seeond indieator 
of quality then is the eoverage of the farm population 
by the CML. Other indie ators of quality relate to the 
aeeuraey and eompleteness of the data, and the 
validity of the procedures used in processing the 
data. 
In some cases, NASS was able to produce measures 
of quality - such as the response rate to the data 
collection, the coverage of the census mail list, and 
the variability of the final adjusted estimates. In 
other cases, measures were not produced but 
descriptions of procedures that NASS used to reduce 
errors from the procedures were subsequently 
provided. 
Census Response Rate 
The response rate is one indicator of the quality of a 
data collection. It is generally assumed that if a 
response rate is close to a full participation level of 
100 percent, the potential for nonresponse bias is 
small, although this has been questioned recently in 
the literature. Because the CML contains both farm 
and nonfarm records, the response rate is an 
indicator of replying to the census data collection 
effort, but does not reflect whether those responding 
met the farm definition. The response rate for the 
2012 Census of Agriculture CML is 80.1 percent as 
compared with a response rate of 85.2 percent for the 
2007 Census of Agriculture and 88.0 for the 2002 
Census of Agriculture. 
The 2012 Census of Agriculture response rate used 
the fourth response rate formula from the American 
Association of Public Opinion Research Response 
Rate Standard Definitions manual: 
c r 
RR4 = (100 ) 
C + R + NC + O + Replicated + e{U ) 
where 
Cadj - number of fully and partially completed 
records, excluding replicated records 
R = number of explicit refusals 
NC - number of non-contacted operations 
O - number of other types of nonrespondents 
Replicated - number of replicated records 
U = number of operations of unknown eligibility 
2012 Census of Agriculture 
USDA, National Agricultural Statistics Service 
e(U) = estimated number of operations of unknown 
eligibility assumed to be eligible 
Records were classified into the above variables 
based on the combination of their active status (AS) 
codes, in-scope status, and replication status. Active 
status refers to the eligibility status of records for 
selection on the CML. All replicated records were 
considered to be a form of nonresponse and were 
classified into other nonrespondents; in-scope status 
was considered immaterial. 
Certain active status classifications indicated records 
of unknown agricultural status. These classifications 
included records to be removed from the CML but 
had data from outside sources indicating agricultural 
activity, new records from outside data sources, 
nonrespondents and refusals to the NACS, records 
for regional office handling only, and records with 
Farm Service Agency or Conservation Reserve 
Program data on operations that are not owned by 
the principal operator. These records were stratified 
(grouped) based on their probabilities of being in- 
scope had they responded. The estimated number of 
in-scope nonrespondents was calculated for the hth 
stratum (group) by the following formula: 
f c. . ) 
e(U,,)= u,^ 
I ^ h J 
where 
e(U/j) = estimated number of operations of unknown 
eligibility assumed to be eligible in the hth group 
Cin-scope,h ~ the number of completed and in-scope 
census records in the h\h group 
Ch - the number of completed census records in the 
hth group 
Uh = number of operations of unknown eligibility in 
the hth group 
Census Coverage 
As a side -product of the statistical adjustment used 
to account for undercoverage, nonresponse of farms 
on the CML, and misclassification of responses to 
the census, the proportion of the adjustments due to 
each of those factors can be derived. The percentages 
of final census estimates due to adjustments for 
APPENDIX A A- 15 
