Methods for variance estimation under random hot deck imputation in business surveys

When the imputed values are treated as if they were observed the precision of the estimates is generally overstated. In the paper three variance methods under imputatation are taken into account. Two of them are the well known bootstrap and Multiple Imputation. The third is a new method based on grouped jackknife easy to implement, not computer intensive and suitable when random hot deck imputation is performed. A simulative comparison on real business data has been carried out. The findings show that the proposed method has good performances with respect to the other ones.