被访者驱动抽样在农民工调查中的应用:实践与评估

Application of Respondent-Driven Sampling in the Survey of Chinese Rural-to-urban Migrant Workers:Practice and evaluation

针对长期以来农民工调查缺乏抽样框的困境,本研究引入被访者驱动抽样(Respondent-Driven Sampling)对全国7个省13个城市(区)农民工进行了较大规模的问卷调查。本次调查针对中国农民工的实际情况,采用非固定点调查法,并对原有方法有所改进。本文是对RDS在农民工调查中的效果评估。数据分析结果表明,此次对农民工的RDS抽样所获样本达到了均衡,且在较短招募批次内快速收敛,所获样本代表性较好。本文认为,RDS抽样虽然不是对目标群体的完全随机抽样,但可以在努力改进和严格执行抽样程序的基础上获得一个渐进无偏的样本,并对样本偏差做出估计。RDS可以作为农民工及其他缺乏抽样框群体抽样的有效的替代性方法。

As there has been a long-term difficulty that researchers do not have proper sampling frame in the survey of Chinese rural-to-urban migrant workers,this study adopted the Respondent-Driven Sampling(RDS) and conducted a large-scale questionnaire survey in 13 cities(districts),7 provinces in China.Based on the Chinese rural-to-urban migrant workers' real situation,this study made use of the unfixed-sites survey and improved the original method.This study is also an evaluation of the application of RDS used in the survey of migrant workers.The results show that samples obtained by the RDS in this study reached equilibrium and converged rapidly in a few of recruiting batches.Thus,samples in this study have good representativeness.This paper argues that although RDS is not a completely random sampling method for the target groups,researchers can obtain an asymptotic unbiased sample through improvement and strict sampling procedure,and the bias of the sample can also be estimated.RDS can be used as an effective alternative approach for the survey of migrant workers and other groups which do not have a clear sampling frame.