Data collected in a population census is usually evaluated by a post enumeration survey, the purpose of which is to mainly indicate the extent of net omission. Other errors that are generated during census data collection are, however, not evaluated. As the census figures are often used without critically examining the data, an analysis of the process by which data are obtained can identify the errors generated. In this paper a framework for analysing measurement error is presented. Based on the framework, the results of a survey of 1981 Census enumerators are discussed. This exploratory survey is focused on the primary demographic data collected in the census using two individual (universal and sample) forms. While some errors are committed by enumerators, other errors occur independent of enumerator characteristics.