[Limdep Nlogit List] parameter varaince in RPL correction

[Brett Smith]

Dear all,

 

Unfortunately, I misread the output. It would seem that ;rpl=var and
;hfn=var are both performing the task. That is estimating the means as
linear functions of var. I guess my question becomes a little simpler:

 

Does anyone know of a way to estimate the variance (spread) of a random
parameter as a function of another variable in RPL? 

 

Regards,

 

Brett

 

  _____  

From: Brett Smith [mailto:Brett.Smith at uwa.edu.au] 
Sent: Wednesday, 4 October 2006 12:05 PM
To: 'limdep at limdep.itls.usyd.edu.au'
Subject: parameter varaince in RPL

 

Dear all,

 

I am aware that for mixed logit it is possible to estimate the means of
random parameters as functions of individual characteristics or context
variables using ;rpl=var and (after some trial and error) it is possible to
estimate the variance (spread) as a function of a variable using ;hfn=var.
However, in attempting to include both specifications it seems that the
heterogeneity (;rpl=var) overrides the heteroscedacity (;hfn=var). Has
anyone managed to include both specifications? Is there a work around in
LimDep? Or am I being a little ambitious?

 

Thanks in advance,

 

Brett.