[Limdep Nlogit List] repost with adapted tables, sorry for inconvenience

Koetse, M.J.J. (Mark) mkoetse at feweb.vu.nl
Wed Nov 8 00:55:29 EST 2006


L.S.,

Mr. Feinberg, thanks very much for your early response to our question.
However, our main problem is not that we do not have the information on
unselected choices, but really is on the set up of the data. Below we
try to describe our problem in more detail.

Our data set is of the following format 

Individuals   Choice    Temperature   Rain   	Income   	Distance

1		  2		3		  1		5
3.5	
2		  3		4		  1		4
6	
3		  4		1		  1		5
10	
4		  3		4		  2		6
5	
5		  5		6		  4		3
20	
6		  6		7		  5		4
8	
7		  7		9		  1		5
25	
8		  1		10		  1		1
1	
9		  4		2		  1		2
15	

Choice, Temperature, rain, income are categorical variables, and
distance is measured in kilometers. We want to estimate a multinomial
logit model for this data set format. We assume that "temperature" and
"rain" affect different mode choice to a different extent. Furthermore,
each individual has to make choice among "seven" choices. As far as we
can see, the data set-up in Limdep needs the following format for using
data for discrete models (we outlined individual 1 and 2):

Individuals	Alternatives Choice Temperature Rain Income Distance	
1		1		 0	  3		  1	 5
3.5	
1		2		 1	  3		  1	 5
3.5	
1		3		 0	  3		  1	 5
3.5	
1		4		 0	  3		  1	 5
3.5	
1		5		 0	  3		  1	 5
3.5	
1		6		 0	  3		  1	 5
3.5	
1		7		 0	  3		  1	 5
3.5	
2		1		 0	  4		  1	 4
6	
2		2		 0	  4		  1	 4
6	
2		3		 1	  4		  1	 4
6	
2		4		 0	  4		  1	 4
6	
2		5		 0	  4 		  1	 4
6	
2		6		 0	  4		  1	 4
6	
2		7		 0	  4		  1	 4
6	

And this setup continues until we have 9*7=63 lines (7 lines per
individual) for the same data format given above. 

Now the questions are:

1-	Can we Use directly the first data format for estimation of
multinomial logit? Or it is necessary for using LIMDEP to have our data
in the second format? 
2-	If Limdep can not handle the data in the first format. Is there
any alternative available for it because our data set is very big and it
will take lot of time and efforts to convert it to the second format?
3-	And even if we convert data to the second format would it not be
anything additional but to reproduce the selected choices values again
for all unselected choices?

Again, any help is appreciated.

Kind regards,

Mark Koetse and Muhammad Sabir


-----Original Message-----
From: limdep-bounces at limdep.itls.usyd.edu.au
[mailto:limdep-bounces at limdep.itls.usyd.edu.au] On Behalf Of Fred
Feinberg
Sent: 03 November 2006 19:06
To: Limdep and Nlogit Mailing List
Subject: Re: [Limdep Nlogit List] (no subject)

The problem seems more severe than would be addressed by reformatting
the data.  It's not clear from your description whether you're running a
(McFadden type) conditional logit model -- iin which the characteristics
of the choices themselves would be changing from choice occasion to
choice occasion -- or a multinomial logit (where you would obtain
coefficients for characteristics of the decision-maker).  It sounds like
you have elements of both, which is fine.  [Note that programs like SPSS
and SAS, for example, will run multinomial models, but not conditional
ones, and have to be tricked into doing so by, for example, Cox
regression.]

Anyway, if you are trying to get some set of coefficients for the
*attributes* of the choices themselves, as common in transportation and
ubiquitous in marketing (e.g., price changes), then not having those
attributes for the options that were not chosen is a BIG problem. One
possibility is to do some type of (perhaps multiple) imputation.  This
is complicated by your need to assume that the he covariates that were
NOT observed are drawn from the same distribution of those that were.
For example, if we don't observe the price of something you didn't buy,
can we assume it's drawn from the same distribution of the prices that
WERE observed?  That's rarely true.  Usually, you'll need a separate
model for the unobserved outcomes, which is where Heckman-type models
can be useful.

Joe Ibrahim (http://www.hsph.harvard.edu/facres/ibrhm.html) has done a
lot of work on this and related problems, primarily from a Bayesian
perspective, which handles missing data in a natural fashion.  He may
have written something on your specific problem.  A good general
reference is:

Ibrahim JG, Chen MH, and Lipsitz SR. Missing Responses in Generalized
Linear Mixed Models When The Missing Data Mechanism is Nonignorable,
Biometrika, 2001; 88:551-564.

FF


"Koetse, M.J.J. (Mark)" wrote:

> L.S.
>
>
>
> We are doing some research in which we want to estimate a multinomial
> logit model on transport mode choices by individuals. We only have
> information on the choices that individuals make, some characteristics
>
> of the transportation mode chosen, and on individual characteristics.
> We
> do not have any information about the choices that are not selected.
>
>
>
> As far as we can see, the data set-up in Limdep requires information
> about the selected choice as well as the non-selected choices. This
> implies that we need to convert our single line observations (in which
>
> one line represents one individual) to multiple line observations (in
> which one individual is represented by a number of lines that is equal
>
> to the number of alternatives in the choice set). However, since we
> have
> no specific information on these possible non-selected alternatives,
> this would imply that these extra rows contain identical information
> as
> the original row.
>
>
>
> Since our dataset is very large, this procedure would take a lot of
> time. Our question is basically whether it is possible to estimate a
> multinomial logit without applying this data transformation procedure?
>
>
>
>
> Any help is greatly appreciated.
>
>
>
> Kind regards,
>
> Mark Koetse and Muhammad Sabir
>
>
>
> Department of Spatial Economics
>
> Vrije Universiteit Amsterdam
>
> De Boelelaan 1105
>
> 1081 HV Amsterdam
>
>
>
> _______________________________________________
> Limdep site list
> Limdep at limdep.itls.usyd.edu.au
> http://limdep.itls.usyd.edu.au
_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au
_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au



More information about the Limdep mailing list