Evaluation of data editing procedures: Results of a simulation approach

Abstract

The impact of data editing on data quality is a theoretical and practical matter, studied by the Joint Group on Data Editing of the Statistical Computing Project Phase 2. In this context this paper focuses on the difference between Imputation at the Record Level and at the Weight Adjustment at the Aggregated Level. After describing two experiments carryied out using a large and complex labour survey, it is stated that redundancy is necessary for record level imputation and that two types of variables, flow variables and semantic variables, affect differently the edit process.