Who Does Dot Respond to the Agricultural Resource Management Survey and Does It Matter?
研究了美国农业资源管理调查中三分之一的样本农场不回应的问题，利用农业普查数据发现，尽管大农场回应率低，但基于回应子样本的计量估计系数仍在全样本置信区间内，表明非回应偏差不太可能影响结论。
The Agricultural Resource Management Survey is the primary annual source of information on U.S. farms, but in a typical year one‐third of sampled farms do not respond. We use Census of Agriculture data to study nonresponse to the survey and how it affects estimates in two econometric models. Despite larger farms responding less, the coefficients estimated from the respondent subsample always fall inside confidence intervals based on draws from the full sample of respondents and nonrespondents. Although nonresponse bias can vary by application, the findings suggest that bias is unlikely to undermine conclusions based on econometrics using respondent data.