Welcome to Axel Michaud's Site

Use linear regression model Y N P I as an example where Y is dependent variable, N P and I are independent variables

Breusch-Pagan Test

Purpose: tests for heteroskedasticity

Method 1:

1. regress Y N P I
2. predict resid, residual
3. generate residsquared = resid^2
4. regress residsquared N P I

Method 2:

1. estat hettest

White Test

Purpose: tests whether variance of errors is constant

Method 1:

1. regress Y N P I
2. predict resid, residual
3. generate residsq = resid^2
4. generate Nsq = N^2
5. generate Psq = P^2
6. generate Isq = I^2
7. generate N_P = N*P
8. generate N_I = N*I
9. generate I_P = I*P
10. regress residsq N P I Nsq Psq Isq N_P N_I I_P

Method 2:

1. estat imtest, white

Heteroskedasticity-Consistent Standard Errors

Method:

1. regress Y N P I
2. regress Y N P I, robust

Durbin-Watson Test

Method:

1. regress Y N P I
2. estat dwatson