Evaluating estimation quality after automatic statistical data editing processes

Abstract

Automatic statistical data editing (ASDE) is a process consisting of a combination of edit steps that correct errors in the data in order to obtain more accurate estimates. The classic estimation method of the reliability of an estimate does not take the effects of ASDE into account. The classic method to estimate the standard error only takes uncertainty due to sampling variance into account. However, because of stochastic components of some of the edit steps, ASDE is likely to add to the reliability of an estimate as well. The aim of this research was to provide an adequate measure for the reliability for analyses involving ASDE. Results from the simulation study show that measuring the standard error with a bootstrap approach provides a better estimation of the estimates reliability compared to classical estimation, because it takes the uncertainty due to the stochastic components into account. In addition, by investigating the standard error over the different edit steps, the bootstrap approach can be used to obtain information about the effects and quality of the edit steps.