[Limdep Nlogit List] Latent Class Respondent Identification
William Greene
wgreene at stern.nyu.edu
Sat Aug 9 10:16:07 EST 2008
Phil: If it were possible to find out which respondent
belongs to which segment, then
(1) The classes would not be latent and
(2) It would make no sense to fit the model as an LCM. You
would just separate the observations into the segments to begin
with and fit separate models for each.
The closest you can get is a best guess of the class based on
the probabilities, which you get with ;CLASS = variable in your
model command. However, this just makes the class assignments
based on the largest posterior probability; it is not definitive.
/Bill Greene
----- Original Message -----
From: "Phil James" <pasj500 at york.ac.uk>
To: "Limdep and Nlogit Mailing List" <limdep at limdep.itls.usyd.edu.au>
Sent: Friday, August 8, 2008 10:33:14 AM (GMT-0500) America/New_York
Subject: [Limdep Nlogit List] Latent Class Respondent Identification
Bill and Others,
I have a discrete choice latent class model with a class probability model.
I am now interested in investigating the, for example, mean age of each
segment. Is there anyway using NLOGIT that I can find out which respondent
belongs to which segment, which will allow me to do this?
Many thanks in anticipation
Phil James
- - - - - - - - - - - - - - - - - -
Philip James
University of York
Heslington
York
YO10 5DD
UK
_______________________________________________
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
More information about the Limdep
mailing list