Bias in Chinese Economic Statistics: The Case of the Typical Example Investigation

Publisher Website
Google Scholar
Abstract
A large number of statistics on the Chinese economy are generated through complete enumeration. Essentially the entire rural population, for example, is organized into agricultural collectives or state farms, and these units are required to submit standardized reports on a range of economic and social activity. They report on several categories of production, on financial flows, and on matters such as births and deaths. There are, of course, areas of interest to Chinese planners and social scientists, the data for which are not routinely compiled at local levels, e.g. data on family consumption patterns. In China, as elsewhere, sample surveys are used to fill such information gaps. Such surveys are commonly conducted using the technique of “typical example investigation” (dianxing diaocha). While well established in China, this survey technique, which might be characterized as stratified non-random sampling, is increasingly challenged as inadequate to the needs of economic science. One recently published large scale typical example investigation generated some statistics for which are available population parameters derived from complete enumeration. Comparison of the results of the typical example investigation and the complete enumeration reveal a strong bias in the sample. The bias is shown to result from the role of social statistics in China, and provides a clear illustration of the contradictions in that role.

This publication has 2 references indexed in Scilit: