A New method for estimating missing data in the multiple linear regression model
Main Article Content
Abstract
the current study aimed to introduce a new method (Slopes Mean) to handle the missing data in the multiple linear regression model and compare it with the Mean Imputation method, Regression Imputation, the (EM) method, and the (FIML) method.
The proposed method is depend on the calculation of the absolute slope between each of the two points in the regression model and then taking the arithmetic mean and compensating it instead of the missing values and comparing the results of this proposed method with the results of other methods that used to handle the missing data using the criteria (MSE, R2 , Adjusted R2 , VIF).
Simulation method have been done to compare these methods.
The results indicated that the proposed method (SM) is the second best way after the EM method because it contains less MSE than the Mean Imputation method, the regression method, and FIML method.
It also contains a coefficient of determination and a coefficient of adjusted determination the larger than these methods.
The results also indicated that the proposed method is better than the EM method in terms of the VIF where it contains less VIF.