项目无回答的成因与降低其水平的途径

本文基于两个全国概率抽样调查的实证数据,采用序列变量逻辑斯蒂回归和稳健回归的分析方法,探讨了在中国社会科学概率抽样调查中影响项目无回答的主要因素,结果发现受访人的受教育水平、城乡、地域属性和年龄是对无回答水平最稳定、最关键的影响因素。在控制了这些人群特征和题目的难度类型之后,测量指标的信度水平对无回答也具有显著的影响。根据分析结果,本文从实践的角度提出了有关降低项目无回答水平的建议。

The Distribution And Origins Of "Item Nonresponse":Lessons Learned From Chinese Social Science Surveys

Based on the data from two national surveys,this study's analysis using survey ordered logistic regression discovers that education,rural/urban residence,area,and age are the most important and stable factors in explaining the level of item nonresponse in Chinese social science survey.Furthermore,among the people with these core characteristics,controlling the type and degree of difficulty of survey questions,this paper finds that the reliability of measurement has a significant linear relationship with the level of item nonresponse.