[Limdep Nlogit List] discrete choice with small proportion of ones
Stijn Bruyneel
stijn.bruyneel at uhasselt.be
Mon Jul 31 18:01:01 EST 2006
Hi,
There exists something called Rare Events Logit. King and Zeng have
developed the RELOGIT program for estimating such rare events. See
http://GKing.Harvard.Edu for more info.
Stijn Bruyneel
-----Oorspronkelijk bericht-----
Van: limdep-bounces at limdep.itls.usyd.edu.au
[mailto:limdep-bounces at limdep.itls.usyd.edu.au] Namens Arthur Caplan
Verzonden: vrijdag 28 juli 2006 17:39
Aan: Limdep and Nlogit Mailing List
Onderwerp: RE: [Limdep Nlogit List] discrete choice with small proportion of
ones
Hi Casey,
I know that there are zero inflation models for count distributions.
You might try and find something similar for discrete choice.
Arthur Caplan
-----Original Message-----
From: limdep-bounces at limdep.itls.usyd.edu.au
[mailto:limdep-bounces at limdep.itls.usyd.edu.au] On Behalf Of Casey
Dawkins
Sent: Friday, July 28, 2006 9:14 AM
To: limdep at limdep.itls.usyd.edu.au
Subject: [Limdep Nlogit List] discrete choice with small proportion of
ones
Hello,
I am estimating a discrete choice model with a sample
that contains a small proportion of ones (less than 5%
of the sample).
I am looking for references on the econometric issues
associated with such samples. I know that
count-R-square-type measures are unreliable for such
models, because the .50 cutoff is a poor naive
prediction rule. I've also seen somewhere that the
logit model may be inefficient in such cases, but I do
not recall where this result appears.
Any suggestions would be greatly appreciated.
Thanks,
Casey
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
_______________________________________________
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