Le data editing : Définition et principes généraux

Data editing refers to the overall process of checking data to ensure that they are of the quality required for analysis by identifying and correcting errors throughout the statistical process. This document presents the basic principles of data editing as discussed in the literature with an educational purpose. The process involves a series of checks: micro-editing and macro-editing checks. We emphasise the importance of having a data "cleaning" strategy that starts with data collection, using built-in checks, continues with automated or manual checks, and ends with post-collection checks. The chosen strategy will depend on the quality of the data collected and the expected quality in terms of accuracy, which should be defined in advance. The metadata accompanying the final dataset is also crucial. In this document, we do not cover methods for correcting data, such as imputation, because the techniques used to impute a plausible value are the same for an incorrect value that needs to be corrected or for a missing value in the case of non-response. We therefore refer the reader to the specific literature on this topic. The references cited in the bibliography will help the reader to find more information on data editing.