[Limdep Nlogit List] Distribution Selection/identification criteria for RPL models

[William Greene]

Dear M. Dutta: There is no standard test for the mixing distribution in a
mixed model.  It is possible - I would not say certainly, nowever - that the
conditions underlying the Vuong test would apply. If so, then you could use
a Vuong test to do the test. On the other hand, I would be very surprised if
this test had much power for the kind of test you propose here.  In the final
analysis, I would guess that you will end up choosing the model that produces
the most reasonable results.
Regards,
Bill Greene

----- Original Message -----
From: "Santa Dutta" <send2santa83 at gmail.com>
To: limdep at limdep.itls.usyd.edu.au
Sent: Tuesday, February 26, 2008 7:55:21 AM (GMT-0500) America/New_York
Subject: [Limdep Nlogit List] Distribution Selection/identification criteria for RPL models

Respected Sir,


Thank you so much for your suggestions which did clear a number of my
doubts.
I am working on Mixed Logit models, where selectioin of distribution is very
critical for identifying the best model. I jhave found that constrained
models are better compared to unconstrained models. However, selecting the
best distribution(E.g.:normal, uniform, triangular, lognormal etc) is a
problem. Can you please suggest something in this regard?  Atleast if
through some test we can tentatively identify the nature of the most
suitable distribution for getting the best model.

Regards,

S.S.Dutta
_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au


-- 
Professor William Greene
Department of Economics
Stern School of Business
New York University
44 West 4th St., Rm. 7-78
New York, NY   10012
http://www.stern.nyu.edu/~wgreene
212.998.0876