摘要
响应和数据缺失是农业经济调查数据中普遍存在的问题,可以分别采取贝叶斯法和Bootstrap法进行多重插补以完成模型构建。文章通过对柑橘主产区种植户调查的缺失数据的实证分析发现,根据两者处理后的农业经济计量模型都有较好的估计检验效果,贝叶斯法有更显著的检验统计量和更精确的区间估计,而Bootstrap法更易于操作。
Response and data missing are common problems in agricultural economic survey data. Bayesian method and Bootstrap method can be used for multiple imputation to complete the model construction. Based on the empirical analysis of the missing data in the survey of farmers in the main citrus producing areas, the paper finds that the agricultural econometric model after the two treatments has a better estimation and test effect, and the Bayesian method has more significant test statistics and more accurate interval estimates, while Bootstrap method is easier to operate.
作者
熊巍
潘传快
祁春节
Xiong Wei;Pan Chuankuai;Qi Chunjie(College of Economics and Management,Huazhong Agriculture University,Wuhan 430070,China;School of Economics,Wuhan Textile University,Wuhan 430200,China)
出处
《统计与决策》
CSSCI
北大核心
2019年第4期11-15,共5页
Statistics & Decision
基金
国家社会科学基金资助项目(16BJY136)
国家现代农业(柑橘)产业技术体系(MATS)专项经费资助项目(CARS-26-08B)
湖北省教育厅人文社会科学基金资助项目(18Y077)
华中农业大学研究生课程建设项目(2015KJ15)
