UDC 004.02

Authors

A.A. Vorobiev, A.A. Voronetsky

Abstract

The paper considers the possibilities and limitations of the multinomial logistic regression method for the redistribution of undecided respondents. As an alternative, the multiple imputation method is considered. A search is made for the rational parameters of the multiple imputation method for the redistribution of undecided ones, which provide improved forecasting accuracy.

Keywords

redistribution of undecided, multinomial logistic regression, multiple imputations, experiment, IBM SPSS Statistics.