Should I use Breush Pagan (koenker) test if I have a linear form of heteroskedasticity? I read that it models the error variance as σ2i=σ2h(z′iα) where zi is a vector of your independent variables. But if "h" a unknown function, Then, It is a general test?
Is the question, Can you use the B-P test on two-sample data? ?
It is indipendent. I want to use Breush-Pagan test for residuals. But it is a specific test or a general test? Are there some initial condition if you can implement it?
Sorry is it a specific test? Are there some initial conditions to implement it?
I'm not very familiar with the test. But from what I understand you can use it on a linear (regression) model that is fit with ordinary least squares (OLS). (That is, linear regression, ANOVA, ANCOVA, etc. But not generalized linear model or mixed model).
It's implemented in R is a few different packages, and the examples there range from a two-sample test to multiple regression with both factor and numeric terms.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics