One thing you might try is working out the equivalent loglinear model and using Poisson regression. It will be uglier than <insert whatever hideous metaphor you want> if have more than a few covariates, but it might work. Using a flattening constant may get you by the perfect predictions.
See Alan Agresti's Categorical Data Analysis, 2nd Ed. (2002) for how to do this.
-----Original Message-----
From: "Tiemann, Michael" <[email protected]>
To: [email protected]
Sent: 5/21/2008 6:01 AM
Subject: AW: AW: st: maximum number of outcomes in mlogit
>Maarten asked:
>Are you aware that there are already many clasifications and scales of
occupations?
Yes, I am fully aware of that. I work for the federal institute for
vocational education and training.
>I realize that you are not going to like the following statement, but I
don't see much reason to reinvent yet another clasification.
The thing is that existing classifications cannot aggregate occupations
in the way we need them to be aggregated. We need tasks to primarily
define groups of occupations, which nominally the KldB92 and 88 follow.
But they are not consequent: besides tasks there are a number of
different criteria used for classifying. If this wasn't the case we'd be
happy to use these classifications.
>For instance, take a look at -findit isko- and -findit isco- for some
Stata implementations of some existing scales and clasifications.
The thing with isco is that we want to use qualifications (ie skills
level here) not to classify occupations but to find out whether they
might change over time for a given set of occupations.
Michael.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
