19th International CODATA Conference
Category: Economic Development

Reporting and Compensation for Non-sampling Errors

Themba Mohoto (ThembaM@statssa.gov.za
Analysis and Consulting [Q&M], South Africa

There are a number of methods used in household surveys to collect and to analyse data. Inherent in these methods of data collection are errors, which can occur in various forms. Silva, 2003 in his paper “Reporting and Compensation for Non-sampling Errors for Surveys in Developing and Transition Countries: Current Practice and Future challenges” indicated that the lack of a widely accepted unifying theory, of standard methods for compiling and estimating parameters of the non-sampling error components, and of a culture that recognises these errors as important to measure, assess and report about, makes their measurement receive little attention in most surveys carried out by developing or transition countries.  There are two broad categories of errors known and used in the literature and these are: Sampling and Non-sampling errors. A great deal of research by statisticians have gone into the area of sampling errors maybe due to ease of computation. However, non-sampling errors have received less attention. In this paper, we propose some possible solution to dealing with non-sampling errors.