REMEDY OF EFFECTS OF MULTICOLLINEARITY IN MULTIPLE LINEAR REGRESSION MODEL

Gad, M. A.; Hussein, H. F. M.

doi:10.21608/fjard.2019.190382

REMEDY OF EFFECTS OF MULTICOLLINEARITY IN MULTIPLE LINEAR REGRESSION MODEL

Document Type : Research articles.

Authors

Cent. Lab. for Design and Stat. Analysis Res., ARC

10.21608/fjard.2019.190382

Abstract

Assuming that there is no complete linear relationship between the
independent variables in the multiple linear regression models leads to the
multicollinearity problem. Some undesirable effects, according to this problem,
resulting imprecise estimators of the equation, and lower ability of the statistical
model to predict the dependent variable in the future. The study aims at
investigating the nature and causes of the multicollinearity problem, methods of
detection and remedy of it using the Ridge regression method as an effective way
of mitigating the effects of multicollinearity.
The study found that the main effect of the multicollinearity problem is
that it may change the signs, significance and estimators of the explanatory
variables. Thus increasing the value of the coefficient of multiple correlation and
the coefficient of determination at a value greater than their true value.
Consequently the low confidence in the results obtained from the statistical
analysis and the low predictability of the relationship between independent and
dependent variables. The study recommended detecting multicollinearity and
identify its causes, and remedy of it using Ridge regression approach.

Main Subjects