Application of Multiple Imputation Method for Missing Data Estimation

Abstract

The existence of missing observation in the data collected particularly in different fields of study cause researchers to make incorrect decisions at analysis stage and in generalizations of the results. Problems and solutions which are possible to be encountered at the estimation stage of missing observations were emphasized in this study. In estimating the missing observations, missing observations were assumed to be missing at random and Markov Chain Monte Carlo technique and multiple imputation method were applied. Consequently, results of the multiple imputation performed after data set was logarithmically transformed produced the closest result to the original data.