nonresponse, and miselassifieation. One model 
estimated the probability of an agrieultural operation 
with unresolved farm status being a farm. The 
remaining four models estimated the probability of 
eoverage, response, and eorreet elassifieation of 
farms and of nonfarms. Eaeh model was fit 
independently by two people. For some models, both 
statistieians obtained the same model. Although the 
eovariates in the two seleeted models differed some 
for the other logistie models, the estimated 
probabilities were similar, but not identieal. The 
reported standard errors aeeount for the variability in 
the parameter estimates of the seleeted models, but 
not for the additional variation due to model 
uneertainty. They also do not aeeount for any bias 
assoeiated with a model. 
2012 Census of Agriculture APPENDIX A A - 19 
USDA, National Agricultural Statistics Service 
