EViews.com

Hello,

Supose you have a tobit equation named tobit and you want the Expected values.

We saw there is a command : tobit.fit yhat (yhat name of fitted residuals ) ,
but also a command tobit.fit(i) yhat ( yhat again the name )

Both give VERY different estimates of yhat..

So.. What is the difference when you add the i between the brackets? It's poorly explained in the manual as "taking into account an index" , but I don't understand this...

Which is the correct method and what is the difference?

Thanks for the help :)

The "i" option produces the fitted index value, or "xb" which may or may not differ from the expected value of the observed response. Note that the fitted index is the expected value of the latent response variable.

I'm not certain why you feel it is poorly documented. We write:

"i - Compute the fitted value of the index. Only for binary, censored and count models"

and offer links to the discussion of those models which describe in detail what we mean by the index.

Thank you for your reply. I understand it now :)

Well .. i really didn't understand "i - Compute the fitted value of the index. Only for binary, censored and count models".
Actually I still don't understand this phrase.. the index?
But because of your explanation of xb ( which can differ from E(y|x) in for example a tobit/probit , I now understand.