Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Limited dependent variable


From   Joseph Coveney <jcoveney@bigplanet.com>
To   Statalist <statalist@hsphsun2.harvard.edu>
Subject   Re: st: Limited dependent variable
Date   Mon, 14 Aug 2006 12:38:21 +0900

Neil wrote:

I am working with panel data and have a dependent variable that can take on
3 possible values of 0, 0.5 or 1. The dependent variable represents a wins
percentage with 1 representing a win, 0.5 representing a tie and 0
reflecting a loss. I wish to regress this wins percentage variable on
several explanatory variables.

>From reading through the FAQs and stata help facilities my understanding is
that this falls within the domain of a generalized linear model. Is it
correct to use -xtgee- and if so what option should I choose for the family
and link choices? Or should I be adopting a different approach?

--------------------------------------------------------------------------------

The wins percentage sounds as if it can be construed as an ordered category.
If that's the case, then you might be better off using ordinal logistic
regression or ordered probit regression.  (-xtgee- doesn't readily do
ordered categories.)  One such command would be something like -ologit
wins_percentage expanatory_variable1 explanatory_variable2 . . .,
cluster(id)-.  -gllamm . . ., i(id) family(binomial) link(ologit)- is an
alternative to consider.  They fit different models.

(If you're trying to construct some kind of league table, then there are
undoubtedly better approaches than what I can suggest.)

Joseph Coveney


*
*   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/



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