政治学研究中的抽样调查:难点、问题与方法创新  

Survey Research in Political Study: Challenges,Problems and Methodological Innovation

作　　者：严洁[1] Yan Jie

机构地区：[1]北京大学政府管理学院,北京市100871

出　　处：《政治学研究》2018年第3期17-28,125,共13页CASS Journal of Political Science

基　　金：国家自然科学基金项目"并行数据与调查数据质量"(71171004)的研究成果

摘　　要：政治学抽样调查面临概念抽象、复杂,难以测量,理论假设中的关系结构复杂,不得不较多依赖面访式概率抽样调查的难点。受这些难点所限,一些调查中发生了概念不清、社会期许偏差、评价参照系偏差、覆盖偏差、无回答偏差,以及抽样成本高昂和无应答率居高不下等问题。针对这些难点和问题,学者们利用列举实验法、随机化回答技术、虚拟情境锚定法来解决社会期许偏差和评价参照系偏差问题;利用地址抽样来解决覆盖偏差问题,以空间单元格和夜间灯光亮度来降低高昂的抽样成本;以并行数据的应用来降低访员效应,处理无应答,构建应答代表性指标,以此提高调查和统计推断的质量。这些创新方法为政治学抽样调查的精准化积累了有益经验。It is difficult to measure the abstract,complicated,and latent political concepts in survey research.It is well-known that there is complex relationship structure in political theory,which brings several fatal challenges for face to face interview on the randomly selected samples.Generally,there are several important problems in the survey study,such as unclear defined concept,social desirability bias,incomparability,coverage bias,nonresponse bias,and heavy cost.In order to overcome these problems,social scientists have employed various methods such as list experiment,randomization response technique,and anchoring vignettes to solve the social desirability bias and incomparability problem.In addition,address sampling,GPS assisted spatial sampling and nighttime brightness were employed to solve the coverage bias and reduce the sampling cost.Furthermore,the parallel data is used to solve the interviewer s effect and nonresponse through generating the representative indicator of response to improve the quality of descriptive and statistical inference.To be summed,these innovative methods provide survey based political study a valuable experience for producing precise knowledge.

关 键 词：调查 抽样 列举实验 虚拟情境 并行数据 

分 类 号：D0[政治法律—政治学]