With The Help Of CGSS Floating Population Sampling Design

Sampling is an important composition in empirical study in sociology, and it is connected to the data collection and quality of survey. The floating population sampling has puzzled many researchers, and some researchers solve this problem by using qualitative methods to avoid sampling problems, or by use of non-probability sampling to obtain research information. With the limitations of methods using the social network and the living condition of floating population, which do not depend on the sampling frame, we choose to continue to rely on the mode of sampling frame. In order to ensure sampling accuracy and reduce spending, in this paper Ⅰ will decompose the floating population sampling into residential and non-residential. We can use CGSS to complete the design of the residential floating population,and then take advantage of the dual sampling frame to complete the floating population of the non-residential, which is divided into a cluster sampling based on geographical frame and a simple random sampling based on the list, respectively. Then we’ll give the estimation and variance estimation of the residential and non-resident floating population, and merge the total estimation at last. We show how to do a floating population design by the aid of large-scale social surveys to ensure sampling accuracy while saving investigative resources.