units, the square root of the variance is computed to 
obtain a random fluctuation measure that is in the 
same units as the original estimate. This is called the 
standard error (se) of the estimate. The standard error 
can then be divided by the estimate itself to show the 
relative size of the standard error to the estimate. 
This ratio is known as the coefficient of variation. If 
this ratio is small, the estimate is quite precise. If this 
ratio is large, the estimate is imprecise. An estimate 
of 100 with a standard error of 2 would result in a 
relative standard error of .02 or 2 percent. This 
would be a very precise estimate. An estimate of 100 
with a standard error of 30 would result in a relative 
standard error of 30 percent. This might be 
considered to be an imprecise estimate. The idea of 
precision can be made a little more clear by stating 
that if the estimate is 100 with a standard error of 2, 
you could be quite confident that the true population 
value would be in the interval 96 to 104 (within two 
standard errors of the estimate). 
Table A provides statistical precision estimates for 
the number of farms, total sales, wholesale sales, 
retail sales for the United States and for each state. 
Table B provides statistical precision estimates for 
the total value of sales by size and operations by type 
of crop for the United States. 
A - 6 Appendix A 
2012 Census of Agriculture 
USDA, National Agricultural Statistics Service 
