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: question on perfect prediction with -ice-


From   Maarten buis <maartenbuis@yahoo.co.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: question on perfect prediction with -ice-
Date   Fri, 17 Dec 2010 13:32:00 +0000 (GMT)

--- On Fri, 17/12/10, Laura R. wrote:
> I am currently experimenting with -ice- to impute a
> variable. I get the error message
> 
> perfect prediction detected, using auglogit to impute
> var_x

that is not an error message, it is just a statement of fact.

> My first question is: How can I stop the process, because I
> receive many of these messages and -q- / -quit- command cannot
> interrupt it.

Have you pressed the break key (red circle with white cross)?

> My second question is, how to deal with the perfect
> prediction. I came up with 3 possible solutions, but I am
> hesitant on which would lead to the best results:
> 
> 1. force -ice- to impute the variables with auglogit by
> inserting the option cmd(var_x:auglogit) in the -ice- command

-ice- is already doing that for you. However, I would consider
why you are getting that message. However, I think that often
this is an indication that there is something seriously wrong
with your imputation model. The only way to solve that is to 
know exactly what the problem is. This requires that you start
staring at the data untill you understand where the problem is.
Building a good imputation model is very hard work and very
complex. If this is your first imputation model, then in
general you should expect it take many months or even years of
reading, modeling, checking, reading again, modeling again, etc.
etc.

More importantly I would take Stephen's comments very very
seriously, and forget about -ice- for now and start reading 
the stuff he refered to. This is really his area of expertise,
so what he told you is a fair preview of what you can expect
as comments from reviewers/advisors/other people who can break
you.

-- Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   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