Comparison between the list wise deletion method and the FULL information maximum likelihood in treating missing values for the variables of multiple regression model

Abstract

The problem of incomplete data for the variables of the multiple regression model is one of the most important problems faced by the researcher in the process of collecting data for the variables of research This is due to several reasons, including controlled reasons such as cost and risk and because of the lack of the possibilities for inspection and other are uncontrolled, such as cases of damage when the tests or loss due to wars or disasters , That the adoption of such incomplete data in the analysis leads to inaccurate results and therefore the loss of data should be addressed by using some methods of treatment that lead to accurate results or near precision. The study dealt with two methods of incomplete data processing for regression model variables. list wise deletion methoed (LD) and FULL information maximum likelihood methoed (FIML).

, The aim of study is to compare these two methods in order to arrive at the best method based on simulations. The results of the simulation experiments revealed that the method of deletion is better than the method of maximizing the full information in the processing of incomplete data, based on the mean error squares of the estimated regression model