The DWS statistics is the number that tests the auto correlation in the errors/residuals from a regression analysis. It’s value will always lie between “0 & 4”" .DWS value of ‘2’ means there is no auto-correlation in the sample,whereas value towards “zero” from ‘2’ means there is +ve auto correlation & “towards 4”" from ‘2’ means -ve auto-correlation. Auto-correlation is a characteristic of data in which the correlation between the values of the same variables is based on related objects. In this data set Auto-correlation for variable ‘final’ it means that the marks of student 2 in final depends on the marks of student 1 in final and that of student 3 in final depends on the marks of student 2 in final and so on. If such auto-correlation occurs we should not build a Linear regression model for our model and we just need to move to time analysis Model.

​

The formula for DWS is [summation {E(i)- E(i-1)^2} / summation {E(i)^2} ] , which is quite complex .Here ei means the expected value based on the regression equation which we have generated . As number of predictors in our model is 2 hence we need to look at the table for DWS with n= 105 & k=2 to obtain dl & du. We need to find these value using the table with alpha=0.05 . We get the value of dl= 1.63 & du = 1.72

​

The durbin Watson table scale for our models ft3 & f23 is as shown below. Both models have 2 predictors so the scale of DWS would be same for both models