Data-driven Imputation Strategies and their Associated Quality Indicators in Economic Surveys
Articles and reports: 11-522-X202500100011Description: The use of modern "data"-driven imputation methods to treat non-response in the context of surveys processed in the Integrated Business Statistics Program at Statistics Canada has previously been explored. It was observed that these methods can lead to high quality imputation and further have the potential to result in broad efficiencies when setting up a particular survey's edit and imputation strategy. However, estimation of the associated total variance, more specifically the component due to imputation, remains a challenge. In this article, two methods for estimation of total variance are proposed and show preliminary results that have motivated us to pursue further research in this area.
Issue Number: 2025001Author(s): Mireuta, Matei; Sivathayalan, Ahalya; Styles, StephenMain Product:Statistics Canada International Symposium Series: Proceedings
