Date:         Thu, 2 May 2002 12:46:07 -0400
Reply-To:     brandon.paris@KODAK.COM
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Brandon L. Paris" <brandon.paris@KODAK.COM>
Subject:      PROC MDC and nested logit models
Content-Type: multipart/alternative;

Hello --

About 15 months ago, I was working on a project that involved
investigating the viability of using the nested logit to model some data.
As part of that exercise, I spent many hours reviewing the capabilities of
SAS as well as just understanding the model construction and nomenclature
from texts such as Ben-Akiva and Lerman (1997).  I actually was able to
identify what a nested logit model would look like for our situation (the
hope was to program this into a usable tool for others to use);
unfortunately, the project was canceled because we didn't have the right
software tools to estimate the models.

We now have purchased the ETS module from SAS and so I've been asked to
re-evaluate the problem, but my notes have since vanished.  I'm trying to
retrace my steps and have been using the examples in the PROC MDC manual
from SAS and have gotten to the point where I can replicate the results in
SAS using the provided code but get stuck on interpreting the results.  My
problem is that I don't recall what the resulting model would look like
(i.e. what the parameterization is), and I am trying to avoid deriving it
again as I did using Ben-Akiva and Lerman.  My question is:

An anyone tell me what the model equation looks like for the example shown
in Figure 8 (page 10) of the PROC MDC chapter in the ETS manual.  The
example comes from Daganzo (1979), which probably has the equation for the
example written outright.  I've been unable to secure a copy of the text
for personal review.  I can then use this for interpretation as well as
attempting to program the example into a usable tool.  Once this is done,
moving to my data should be relatively simple.

Any assistance with this would be greatly appreciated.   And if the
technical writers from SAS are reading, it sometimes helps to show the
actual model being fit in examples, in addition to the more cryptic,
mathematical models.

Brandon L. Paris,
Senior Analyst, Business Research
kodak.com, Eastman Kodak Company
(585) 724-3064 (KNet 22-43064)
brandon.paris@kodak.com
"In God we trust.  All others must bring data."  W. Edwards Deming
[text/html]