Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: clogit and lroc question


From   Steve Samuels <sjsamuels@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: clogit and lroc question
Date   Sat, 28 Aug 2010 20:52:05 -0400

--

Libby Williamson:

• Without seeing your code (which the FAQ request that you show), it's
difficult to say much about it.  One guess is that your problem would
be solved if you write: "predict p if(esample), pu0". This will
restrict the creation of the predicted value to observations that were
used in the previous -clogit- command. Since -lroc- works only after
-logistic-, not -clogit-. You'd have to do the ROC graph yourself with
-roctab-.

• -clogit- assumes an unconditional logistic model in which every
group has its own intercept. However the program does not estimate
that intercept. To make the pu0 prediction, the intercepts are all set
to zero. To my mind, that makes the predictions, and any ROC curves
derived from them, pretty useless.  If there are reasons for thinking
otherwise, I'd like to learn them.

• Instead, you might try -xtmelogit- which will estimate the
intercepts, albeit with some assumptions attached. You could then
estimate the predicted unconditional probabilities and use Roger
Newson's -senspec- (from SSC) and -roctab- to get ROC curves.

Steve

Steven J. Samuels
sjsamuels@gmail.com
18 Cantine's Island
Saugerties NY 12477
USA
Voice: 845-246-0774
Fax:    206-202-4783

On Sat, Aug 28, 2010 at 4:55 PM, Libby Williamson <libuvm02@gmail.com> wrote:
> To whom it may concern at stata listserve
> I am using clogit to run a series of fixed-effects models.  I am
> running these models on 4 different populations (within the dataset)
> and each populations runs an additional 4 times for each season (four
> seasons).  I would like to use the predict pu0 and lroc (or roctab)
> but am having trouble with the syntax.  When I list the predict and
> roctab commands after the model, it wants to run on all the data and
> not to the  model set that I just ran with clogit.  Any thoughts or
> suggestions?
> Libby
>
> --
> Libby Williamson
> Graduate Student (MSc)
> Natural Resources and Environmental Studies
> University of Northern British Columbia
> 3333 University Way
> Prince George, BC
> V2N 4Z9
> (250) 960-6739
> email: williame@unbc.ca
>
> "In the wilderness is the preservation of the world"  --Henry David Thoreau
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index