The Remedy Of Nonresponse Missing Data

Before 1990s, nonresponse phenomenon in survey sampling was not very prevalent in our country. There were two main reason. On the one hand, the content hardly related to sensitive questions, such as tax dodging, cheating etc; on the other hand, the survey organization was mainly set up by the executive unit of our country, so it was highly trustful and survey samplings could be finished favoring. However, with the reformation coming since 1990s, some private survey organizations are established. Then many surveys treat of sensitive questions and the nonresponse phenomenon becomes more and more, which affects the precision of the estimate. Now, there is not enough systemic theory study and effective managing methods, so almost all organizations use the stochastic substitute or the omitting missing data method. But its reliability will be reduced. As a result, the article will do some pandect and research under lots of information from inland and outland. First, it sets forth the concepts and the features concerned nonresponse basing on the related data of both inside and outside the country, and analyses the statistical effects of nonresponse under the deterministic view and stochastic view, further advances the settlement way. Second, according to the two aspects of raising response rate and controlling the respondent-nonrespondent differences, it selected some effective and practical method dealing with nonresponse to make a detail assessment, basing on the outcome of study of both inside and outside the country. Final, according to the above and the conditions of our country, I choice and commend some methods that is fit for the conditions of our country, and put forward a new way. At last, preliminarily building a general pattern of methods dealing with nonresponse.