Hi!
I am doing an analysis with the continuous dependent variable "inattention". This is a sum variable ranging from 9-36. The variable is very skewed, but follows a normal distrubution when log transformed. I have read that several people recommend using Poisson with the vce(robust) option in these cases, instead of log transforming the dependent variable and do a linear regression. However, several of my models contain interaction terms, and I want to assess the significance of these interactions. Normally, I would use LR-tests. However, how do you compare models with the vce(robust) option (besides the AIC and BIC)?
Best,
Kjell Weyde
I am doing an analysis with the continuous dependent variable "inattention". This is a sum variable ranging from 9-36. The variable is very skewed, but follows a normal distrubution when log transformed. I have read that several people recommend using Poisson with the vce(robust) option in these cases, instead of log transforming the dependent variable and do a linear regression. However, several of my models contain interaction terms, and I want to assess the significance of these interactions. Normally, I would use LR-tests. However, how do you compare models with the vce(robust) option (besides the AIC and BIC)?
Best,
Kjell Weyde
Comment