From ssingh10 at unl.edu Tue Oct 3 17:49:35 2017
From: ssingh10 at unl.edu (Sunil Kumar Singh)
Date: Tue, 3 Oct 2017 06:49:35 +0000
Subject: [Limdep Nlogit List] Latent Class Models
Message-ID:
Hi -
I am interested in running a probit with latent class. By default, the LCM command provides the output for all the variables in RHS. I am interested to see if I can do some sort of variable selection that tells me for which variables in the RHS the LCM should be run and which should be merely used as a covariate/control. A sample of my code is as follows -
Skip $
probit; lhs= Y
; rhs= one,X1, X2, X3, X4
; LCM = X5
; pts = 2$
Here I am asking the model for 2 latent classes based on X5. But I am interested to check if I should include all X1-X4 variables in rhs or I can truncate that list to say X1, X2 only as hypothetically X3 and X4 don't add any value to the model and I merely control for them. I want to figure out how to do this.
Thanks,
Sunil
From david.hensher at sydney.edu.au Tue Oct 3 18:01:30 2017
From: david.hensher at sydney.edu.au (David Hensher)
Date: Tue, 3 Oct 2017 07:01:30 +0000
Subject: [Limdep Nlogit List] Latent Class Models
In-Reply-To:
References:
Message-ID: <820C00F7-F601-48E6-97B0-A65FFADB6D3C@sydney.edu.au>
You should look at the attribute processing literature and examples we have used to test whether some attributes are ignored. See the second edition of Applied Choice Analysis book for details and Nlogit 6 manual.
One possible approach if the 2^k model
David
Sent from my iPhone
0418 433 057
David A Hensher
Note: hgroup at optusnet.com.au has been cancelled so instead use
hgroup at hensher.com.au David.hensher at bigpond.com
David.hensher at sydney.edu.au
These emails are linked so use one only
On 3 Oct 2017, at 5:50 pm, Sunil Kumar Singh via Limdep > wrote:
Hi -
I am interested in running a probit with latent class. By default, the LCM command provides the output for all the variables in RHS. I am interested to see if I can do some sort of variable selection that tells me for which variables in the RHS the LCM should be run and which should be merely used as a covariate/control. A sample of my code is as follows -
Skip $
probit; lhs= Y
; rhs= one,X1, X2, X3, X4
; LCM = X5
; pts = 2$
Here I am asking the model for 2 latent classes based on X5. But I am interested to check if I should include all X1-X4 variables in rhs or I can truncate that list to say X1, X2 only as hypothetically X3 and X4 don't add any value to the model and I merely control for them. I want to figure out how to do this.
Thanks,
Sunil
_______________________________________________
Limdep site list
Limdep at mailman.sydney.edu.au
http://limdep.itls.usyd.edu.au
From Warwick.Wainwright at sruc.ac.uk Tue Oct 3 23:25:10 2017
From: Warwick.Wainwright at sruc.ac.uk (Warwick Wainwright)
Date: Tue, 3 Oct 2017 12:25:10 +0000
Subject: [Limdep Nlogit List] RPL and GMX logit modelling
Message-ID: <3FEA9BD9EBA7824F8D51B2B2AA764EB14727A8CA@DAGAMB02.internal.domain.loc>
Dear Nlgoit community,
I have been modelling using an RPL model (one with covariates included, one without) and have done the same using a GMX logit model (one with covariates included, one without) to determine WTP in WTP space. When I do the modelling the results I get from the models without covariates are similar (i.e. by significance and sign). But when I contrast the models that have used covariates the RPL and GMX give totally different results. Just wondering why this might be? I am a bit stuck on what I should do next....
Many thanks,
Warwick
Please don't print this e-mail unless you really need to.
This e-mail message is confidential to the intended recipient at the email address to which it has been addressed. If the message has been received by you in error, it may not be disclosed to or used by anyone other than the intended addressee, nor may it be copied in any way. If it is not intended for you please inform us, immediately, then delete it from your system. If the content is not about the business of the organisation then the message is not from us nor is it sanctioned by us. Anything in this e-mail or its attachments which does not relate to SRUC's or SAC Commercial Limited's official business is neither given nor endorsed by SRUC or SAC Commercial Limited.
SRUC
A Charitable company limited by guarantee, Scottish Charity Number: SC003712.
Registered in Scotland, Company Number: SC103046 - Registered Office: Peter Wilson Building, King?s Buildings, West Mains Road, Edinburgh EH9 3JG
SAC Commercial Limited, an SRUC company
Registered in Scotland, Company Number: SC148684 - Registered Office: Peter Wilson Building, King?s Buildings, West Mains Road, Edinburgh EH9 3JG
From michael.danne at agr.uni-goettingen.de Thu Oct 12 00:12:31 2017
From: michael.danne at agr.uni-goettingen.de (Danne, Michael)
Date: Wed, 11 Oct 2017 13:12:31 +0000
Subject: [Limdep Nlogit List] Poe Test
Message-ID: <9D37660B579CDB49A3614369BB58F89CFB1EA625@um-excdag-a05.um.gwdg.de>
Dear all,
I`m struggeling in understanding the difference between the Poe test and the Kolmogorov-smirnoff test for comparing two distributions.
Is there a way to compute Poe in NLOGIT?
Kind regards,
Michael
From wgreene at stern.nyu.edu Thu Oct 12 02:06:08 2017
From: wgreene at stern.nyu.edu (William Greene)
Date: Wed, 11 Oct 2017 11:06:08 -0400
Subject: [Limdep Nlogit List] Poe Test
In-Reply-To: <9D37660B579CDB49A3614369BB58F89CFB1EA625@um-excdag-a05.um.gwdg.de>
References: <9D37660B579CDB49A3614369BB58F89CFB1EA625@um-excdag-a05.um.gwdg.de>
Message-ID:
Michael. Assuming you are using version 6 (I'm not sure this is available
in
version 5), you can use
CALC ; List ; Poe(vector1,vector2) $
This produces the statistic based on the full convolution. Vectors must be
defined
with a MATRIX command. If x1 and x2 are two variables, use
MATRIX ; vector1 = x1 ; vector2 = x2 $
then CALC Poe. The test can be based on the difference of the vectors if
they
are the same length. Use
CALC ; List ; Poe(vector1,vector2,1) $
Regards
Bill Greene
On Wed, Oct 11, 2017 at 9:12 AM, Danne, Michael <
michael.danne at agr.uni-goettingen.de> wrote:
> Dear all,
>
> I`m struggeling in understanding the difference between the Poe test and
> the Kolmogorov-smirnoff test for comparing two distributions.
> Is there a way to compute Poe in NLOGIT?
>
> Kind regards,
> Michael
>
>
> _______________________________________________
> Limdep site list
> Limdep at mailman.sydney.edu.au
> http://limdep.itls.usyd.edu.au
>
>
--
William Greene
Department of Economics
Stern School of Business, New York University
44 West 4 St., 7-90
New York, NY, 10012
URL: https://protect-au.mimecast.com/s/aYDOBJinN6z1H2?domain=people.stern.nyu.edu
Email: wgreene at stern.nyu.edu
Ph. +1.212.998.0876
Editor in Chief: Journal of Productivity Analysis
Editor in Chief: Foundations and Trends in Econometrics
Associate Editor: Economics Letters
Associate Editor: Journal of Business and Economic Statistics
Associate Editor: Journal of Choice Modeling
From moja_afzali at yahoo.com Sun Oct 15 19:26:18 2017
From: moja_afzali at yahoo.com (mohammad Javad Afzali)
Date: Sun, 15 Oct 2017 08:26:18 +0000 (UTC)
Subject: [Limdep Nlogit List] need to sfa commands
References: <954781771.126259.1508055978375.ref@mail.yahoo.com>
Message-ID: <954781771.126259.1508055978375@mail.yahoo.com>
I am a Ph.D. student in Iran and in part of my thesis I need to measure the technical efficiency using Limdep software. Although I have read about this software, but I do not have the commands required to run the model, and no one has the skills to do this. number of variables are 5 with a dummy variable in three years (as unbalanced form) and for 3 function included?Cobb Douglas,?Translog and?Transcendental. If possible, send the commands required to run the software. I need to lim files related to them. I will be very grateful if you send me the software commands.ThanksS.M.J. Afzali?==================================== Seied Mohammad Javad Afzali Iran- Ahwaz Shahid Chamran University Faculty of Agriculture Dept. of Agricultural Machinery P. O. Box: 61849-17881 Tel: +98 9166208840 FAX: +98 21 3737357 ====================================
From Charles.Raux at laet.ish-lyon.cnrs.fr Sat Oct 21 21:53:50 2017
From: Charles.Raux at laet.ish-lyon.cnrs.fr (RAUX Charles)
Date: Sat, 21 Oct 2017 10:53:50 +0000
Subject: [Limdep Nlogit List] dstat, values and frequency
Message-ID: <72D6B3FEC474FA40B4981EDD2D672981CD3EF970@CNMB02WVP.core-res.rootcore.local>
Hello,
I have an issue after importing my data. I have 5400 observations with 59 variables including a variable 'Effect' which I know has 7 values (1..7 for each observation, and this has been checked under Excel in the original csv file: 1800 observations for Effect=1, 600 for each other).
After import I have
|-> dstat; rhs=effect; value$
--------+---------------------------------------------------------------------
| Standard Missing
Variable| Mean Deviation Minimum Maximum Cases Values
--------+---------------------------------------------------------------------
EFFECT| 3.333333 2.160447 1.0 7.0 5400 0
__Values|Negative 0 Zero 0 Positive 5400
|Integers 5400 Real = 0 Unique Values 7
--------+---------------------------------------------------------------------
|-> frequency; rhs=effect$
-------------------------------------
Frequency Table for EFFECT
Sample size 5400
-------------------------------------
Sample Sample
Value Frequency Proportion
------------- --------- ----------
EFFECT= 1 1799 .3331
EFFECT= 2 599 .1109
EFFECT= 3 599 .1109
EFFECT= 4 599 .1109
EFFECT= 5 599 .1109
EFFECT= 6 599 .1109
EFFECT= 7 599 .1109
-------------------------------------
This sums to 5393. Why 1799 (and not 1800), then 599 (and not 600) for each other?
CR
From ambi.ganesh at gmail.com Mon Oct 30 14:20:54 2017
From: ambi.ganesh at gmail.com (Ganesh A R)
Date: Mon, 30 Oct 2017 08:50:54 +0530
Subject: [Limdep Nlogit List] How to store the probabilities
Message-ID:
Respected Sir,
I am currently using NLOGIT to create user defined likelihood function for
discrete choice models.
I would like to know how to store the probabilities I obtain from the
MAXIMIZE function since "PROB = " doesn't seem to work in this particular
function.
?
Thanking
?Y?
ou
Yours Sincerely
Ganesh
?PhD student
IIT Madras, India
?
From wgreene at stern.nyu.edu Mon Oct 30 23:12:13 2017
From: wgreene at stern.nyu.edu (William Greene)
Date: Mon, 30 Oct 2017 08:12:13 -0400
Subject: [Limdep Nlogit List] How to store the probabilities
In-Reply-To:
References:
Message-ID:
Ganesh. When you use MAXIMIZE, the variable LOGL_OBS contains the
individual
observations in the maximand. If they are logs of probabilities, you can
just create
exp(logl_obs) to get the probabilities. Otherwise, you will use CREATE
with the estimated
parameters and the data to compute the function you are looking for.
MAXIMIZE does not
know that it is computing logs of probabilities, or probabilities. It
knows only that it is
computing a function. As such, ;PROB=name would not have any meaning in
this context.
/Bill Greene
On Sun, Oct 29, 2017 at 11:20 PM, Ganesh A R via Limdep <
limdep at mailman.sydney.edu.au> wrote:
> Respected Sir,
>
> I am currently using NLOGIT to create user defined likelihood function for
> discrete choice models.
> I would like to know how to store the probabilities I obtain from the
> MAXIMIZE function since "PROB = " doesn't seem to work in this particular
> function.
>
> ?
> Thanking
> ?Y?
> ou
> Yours Sincerely
>
> Ganesh
>
> ?PhD student
> IIT Madras, India
>
> ?
> _______________________________________________
> Limdep site list
> Limdep at mailman.sydney.edu.au
> http://limdep.itls.usyd.edu.au
>
--
William Greene
Department of Economics
Stern School of Business, New York University
44 West 4 St., 7-90
New York, NY, 10012
URL: https://protect-au.mimecast.com/s/W91ABZiQekKkCX?domain=people.stern.nyu.edu
Email: wgreene at stern.nyu.edu
Ph. +1.212.998.0876
Editor in Chief: Journal of Productivity Analysis
Editor in Chief: Foundations and Trends in Econometrics
Associate Editor: Economics Letters
Associate Editor: Journal of Business and Economic Statistics
Associate Editor: Journal of Choice Modeling
From john.c.whitehead at gmail.com Mon Oct 30 23:47:50 2017
From: john.c.whitehead at gmail.com (John C. Whitehead)
Date: Mon, 30 Oct 2017 08:47:50 -0400
Subject: [Limdep Nlogit List] Saving the conditional mean and standard
deviation of attribute coefficients for ANA models
Message-ID:
NLogit list,
I'm trying to estimated an inferred attribute non-attendance model
following Hess and Hensher (2010
). In
that paper they save the conditional mean and standard deviation of
attribute coefficients and use the coefficient of variation to identify
non-attendance. I've spent some time with the NLogit manual and Applied
Choice Analysis but have only managed to figure out how to save the
conditional mean (without the standard deviation) of the coefficient or the
conditional mean and standard deviation for the first k=1 observations,
where there are k=3 choice questions (I'd like to have the mean and
standard deviation for all k=3 questions ... it might be that I just can't
figure out how to expand the k=1 variables to the full k=3 sample).
Thanks for any help!
John Whitehead