{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

Correcting Heteroskedasticity

# Correcting Heteroskedasticity - Step 5 Using all the new...

This preview shows page 1. Sign up to view the full content.

Correcting Heteroskedasticity Step 1: Using the five original independent variables, we estimate a regression model. Step 2: Computing the residual RES(in MINITAB, create a new variable by using feature RESIDUAL in Storage when running regression model) and their squares ut2 RESSQ(create a new variable by squaring RES) Step 3: Create the ln of ut2 (using LOGE(RESSQ) to create new variable RESLN) Step 4: Create all the interaction of all the original variables(create a series of new variables by squaring all original variables and compute the products of them. Example : 3 original variables X1, X2, X3, we will have 6 new variables X12 , X22 , X32 , X1*X2, X1*X3, X2*X3. )
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Step 5: Using all the new variables created in step 3 and 4, estimate a regression model with them as independent variables and the original dependent variable as independent variable. Step 6: Calculate the fitted ln st2 (create a new variables called FIT by using feature FIT in Storage when running regression model in step 5 to create a new variable LNST). Step 7: Calculate st2 by taking the anti log of ln st2 (create new variables STSQ by using EXPO(FIT)) Step 8: Obtain EGLS by dividing each original independent variables with STSQ to create 5 new variables. Step 9: Using the five new variables from step 8, we estimate a regression model....
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online