Author : 1 Chinelo Mercy Igwenagu (Ph.D), 2 Ozlem Gurunlu Alma (Ph.D)

Multicolinearity is a common problem usually encountered in model building using multiple regression analysis. The effect is such that it affects the interpretation and decision as to which model gave a best fit, especially with variations on the measure of goodness of fit R2, t-ratios, variance inflation Factors (VIF) and eigenvalues of the variables used in modeling. Hence, this paper has examined the effect of multicolinearity in modeling climate variables in Nigeria using data collected from the six geo-political Zones in the country. The result of the multiple linear regression analysis revealed a strong effect of multicolinearity among the variables used with VIF >5. To overcome this effect of multicolinearity, principal component analysis was introduced.

The result obtained from the principal component regression analysis of the retained variables yielded a more preferable result; than the result obtained when all the variables were used with VIF value of 1.039. Therefore, this study suggests that, in an effort to overcome the effect of multicolinearity, it is advisable to carry out principal component regression analysis on only the variables retained after carrying out Ordinary Least Square multiple regression analysis.

Affiliation :

1 Department of Industrial Mathematics/Applied Statistics and Demography, Enugu state University of Science and  Technology, Nigeria.  
2 Department of Statistics. Mugla Sitki Kocman University, Turkey

Keywords : Effect, Climate variables, Eigen values, Multicolinearity, and Variance Inflation Factors
Date : Friday ,01 ,August ,2014

Download Full Page

<<< Go Back