Thesis (M.Sc.) - Cairo University - Institute of Statistical Studies and Research - Department of Statistics and Econometrics

Multicollinearity is considered one of the most important problems that poses the linear regression model and which results in many risks in the assumptions of the model and these risks are : ) The difficulty of parameter estimation of society in the linear multiple regression model. ) Increased of variance value of estimator of society in the model of linear multiple regression. ) Reducing the quality of estimating of ordinary least squares of parameter of society in the multiple linear regression model. ) Effects determining the quality of true linear model. This multicollinearity may be total, linked with two variables or more from explanatory variables in the model and may be partial linked with only one variable of the explanatory variables. There are different methods to solve multicollinearity in the model of the linear multiple regression:- 1) There are difficults of the signs of the society parameter which express the relation of explanatory variables to the dependent variable in the model of the linear multiple regression of its true value in the economic theory.2 ) Increased of the value of coefficient of determining when most of variables of the explanatory variables. 3) Increases of the value of variance inflation factor4 ) The difference of the model in the F test from the model in the T test.5 ) The not equal value of the model for explanatory variable in the model of linear multiple regression from its value of the same variable in the model of simple linear multiple regression6 ) The increases of the conditional number from 10

Issued also as CD