L'imputation des données manquantes, la technique de l'imputation multiple, les conséquences sur l'analyse des données: l'enquête 1999 KOF/ETHZ sur l'innovation
Schweizerische Zeitschrift für Volkswirtschaft und Statistik / Revue Suisse d'Economie politique et de Statistique / Swiss Journal of Economics and Statistics, Volume 137, Issue 3, 2001, Pages 301-317
Abstract
The imputation of missing values is generally used to correct the item nonresponse of a survey. For the KOF/ETHZ-Innovation Survey 1999, this problem is managed with a multiple imputation. We give some theoretical backgrounds of the method and illustrate it by some examples. We compare our approach with the one used by Eurostat for the Second Community Innovation Survey (CIS 2).