Dear all,
I'm looking to fit a linear model for my rather large sample (N=~400) where the dependent variable distribution is right-skewed (satisfaction scores, which have that tendency) and the assumption of homoskedasticity is violated.
Now, I do know that log-transforming the DV can help with distribution issues, and WLS regression is a good way to deal with heteroskedasticity. But can I combine the two? Is there a more appropriate way to go about this?
Thank you so much!
I'm looking to fit a linear model for my rather large sample (N=~400) where the dependent variable distribution is right-skewed (satisfaction scores, which have that tendency) and the assumption of homoskedasticity is violated.
Now, I do know that log-transforming the DV can help with distribution issues, and WLS regression is a good way to deal with heteroskedasticity. But can I combine the two? Is there a more appropriate way to go about this?
Thank you so much!