关键词: 抽样调查, 权数, 包含概率, 截取

Abstract: The role of weight is very important in inferring the population from the sample information. It can not only restore the sample, but also adjust the structure of the sample to be the same as the population by using the weighting. So the correct use of weight is the basis of statistical inference. This paper discusses the acquisition process and adjustment process of weight in sampling survey. Weighting may increase the standard deviation of the estimator although it can improve accuracy of estimation. This paper discusses some methods about controlling the weight. We investigate the effect of the method by applying the data from CGSS. The results show that the adjusted weight can not only improve the precision of estimation, but also reduce the effect of weight in sampling inference.

Key words: Sampling Survey, Weight, Inclusion Probability, Trimming