Robust Regression (rreg) , help for noconstant

M

*marie*

Guest
#1
Dear all,

I am confronted with nonnormality in my data (for the first time, therefore I am a beginner in how to manage this). My data is a time series and skewed (right). I read that the stata codes "rreg" or "qreg" can solve for this problem, but not in case of working without a constant (noconstant).
Can anyone please help me how to regress correctly?

(My concrete regression is to explain forecasting errors of unemployment rates with dummys for election periods and I would like to include both dummys (election and nonelection period))
 
#2
Hi Marie,

What type of model are you looking to use?

If its a time series have you considered a GLM or GLMM?

How non-normal is your data?
 
M

*marie*

Guest
#3
Hi,
thanks for your reply.
Even though my data is time series data I am working with OLS, as I do not inlude lagged variables (My dependent variable is defined as a difference),
but to solve the problem I changed the definition of my dummys to include a constant, which make things much easier. =)