小域估计在问卷分割中的应用研究

Research on Small Area Estimation in Split Questionnaire Design

问卷分割设计在设计上能够通过减少问卷长度来减少调查时间,从而提高问卷回答率和提升调查效率,近年来成为国内外政府部门和调研机构用来提升调查数据质量的一种方法。在涉及对样本数据的估计中,通过问卷分割得到的样本在某些域上往往存在样本量很少的情况,采用小域估计方法对问卷分割设计得到的样本进行参数估计,通过理论推导和模拟研究表明,使用小域估计方法对分割问卷得到的数据进行参数估计,从偏差、相对偏差、估计的均方误差和相对有效性等衡量指标上优于多重插补法得到的估计结果,能显著提高统计调查的精度。但是,当研究的目标变量间存在较强的相关关系时,就需要进一步优化该方法从而得到更精确的估计。

Large-scale survey projects often set more and more questions to ask respondents to answer,resulting in an increase in the non-response rate of the questionnaire and a decrease in the quality of the questionnaire,which in turn affects the efficiency of the estimation.The split questionnaire design can reduce the survey time by reducing the length of the questionnaire in design,thereby improving the questionnaire response rate and improving the survey efficiency.In recent years,it has become a method used by some government departments and research institutions at home and abroad to improve the quality of survey data.In the estimation of sample data,the sample obtained by split questionnaire design often has a small sample size in some areas.In this paper,the small area estimation method is used to estimate the parameters of the sample obtained by the split questionnaire design.Through theoretical derivation and simulation study shows that the small area estimation method is used to estimate the parameters of the data obtained from the split questionnaire,which is superior to the estimation results obtained by the multiple imputation method(PMM)on the measurement indicators of deviation,relative deviation,estimated mean square error and relative validity,and could significantly improve the accuracy of statistical surveys.However,when there is a strong correlation between the studied target variables,the method needs to be further optimized to obtain a more accurate estimate.