[Limdep Nlogit List] LCM - Segment Membership

[Phil James]

Bill,

You have successfully answered my questions. Many thanks. Can I add how
valuable I have found the Limdep lists; I have learnt so much from other
people problems and your feedback. Thanks for your time.

Phil James


-----Original Message-----
From: limdep-bounces at limdep.itls.usyd.edu.au
[mailto:limdep-bounces at limdep.itls.usyd.edu.au] On Behalf Of William Greene
Sent: 31 July 2008 16:00
To: Limdep and Nlogit Mailing List
Subject: Re: [Limdep Nlogit List] LCM - Segment Membership

Phil. Dou you mean that the class probabilities are to be functions
of the demographics?  If so, then you might use
;LCM=fe,nf,goody
However, this does not look right for your model, since these three
appear to be the parameter names in one of your utility functions. As
such, I'm not sure what you have in mind by "membership likelihood 
based on the sociodem data."  The estimated prior class probabilities
for your model will be constants, and will be reported in the output.
The posterior values can be saved in the data set using ;classp=name,
but these are functions of the full likelihood, not just the demographics,
that is, all the data used to fit the model.

About ;RST, you can fix a coefficient to any value you wish by placing
that specific value in the list.
;Rst=beta1,beta2,beta2,23.4589
constrains the second and third parameters to equal each other, and
fixes the fourth at 23.4589.

/Bill Greene

----- Original Message -----
From: "Phil James" <pasj500 at york.ac.uk>
To: limdep at limdep.itls.usyd.edu.au
Sent: Thursday, July 31, 2008 9:45:18 AM (GMT-0500) America/Bogota
Subject: [Limdep Nlogit List] LCM - Segment Membership

Dear All,

 

First may I thank you for your response regarding the fixed parameter, I was
wondering the same. 

 

Second, using an LC model in two segments, how do I work out the membership
likelihood based on the sociodem data that I have using limdep?

 

I am using the coding below, with fe, nf and goody being my individual
related socio dem data.

 

NLOGIT

;LHS= CHOICE, CSET, ALTI

;CHOICES=A,B,SQ

;lcm

;pds=6

;pts=2

;MODEL:

    u(A)= WORK*WORK1+START_a*STARTa+start_b*startb+LABOUR*LABOUR1+
GOOD*GOOD1 + BAD_a*bADa + BAD_b*badb

/

    u(B)= WORK*work1+START_a*STARTa+start_b*startb+LABOUR*LABOUR1+
GOOD*GOOD1 + BAD_a*bADa + BAD_b*badb

/

    U(SQ)=ONE+fe*female+nf*non_farm+goody*good_y   $

 

When fixing coefficients I used the rst code as suggested. This I presume
fixes the coefficients to zero as opposed to normalised them? 

 

Many Thanks

 

Phil

 

- - - - - - - - - - - - - - - - - - 

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

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