An evaluation of the one percent clustered sample of the 1990 Census of China

Abstract

We describe and evaluate a one percent clustered sample of the 1990 Census of China, the largest publicly-available micro sample of any Chinese census, using direct inspection as well as comparisons with published data drawn from the complete enumeration. In the absence of official documentation, we elucidate the basis of the clustering; detect duplicate cases; report corrected totals; and make comparisons between the sample data and tabulations based on the complete enumeration at the province and county levels. Although the sample contains several anomalies, we conclude that it is broadly serviceable.