Though the significance levels of the estimated parameters of the sample and that of the sample regression model is tested, it still remains to examine whether the features of the Classical Linear Regression Model are satisfied or not. There are three main factors which an estimated sample regression model must be testified as not being characterized by, in order to be declared as a CLRM, namely, autocorrelation, heteroscedasticity and multicollinearity. The presence of any of the aforementioned features rules out the fact that the variables being considered have a valid relationship or an association between them.
There are different tests to check for the presence of the above three characteristics. In the present paper, autocorrelation will be checked with the help of the Durbin-Watson test, heteroscedasticity will be assessed with the help of White’ s test and the VIF test will take care of the multicollinearity factor. Autocorrelation investigates whether the consecutive estimated residual terms are correlated with each other or not; in case they are, the implication would be that the variables being considered are not abundant to suitably explain variations in the dependent variable and that the effects of some of the highly relevant variables are being captured by the random error term.
Heteroscedasticity, on the other hand, implies that the estimated random error terms are biased in the sense that the magnitude of difference between the estimated and the actual regression model are significantly different for each observation of the independent variable; this results in an overstated estimated Student’ s t-statistic so that the corresponding null hypotheses tends to be rejected even when they are true. Lastly, multicollinearity implies that the independent variables being included in the model correlate with one another so that the explanatory power of the model automatically increases due to the multiplicative nature of the association between the variables; however, in truth, none of the variables individually portray significant explanatory power over the dependent variable being considered.
Please type your essay title, choose your document type, enter your email and we send you essay samples