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: pseudo R-squared


From   Steven Samuels <sjsamuels@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: pseudo R-squared
Date   Mon, 4 Apr 2011 17:16:59 -0400

--

Gao, you must also make sure that all models are estimated on the same set of observations. That means you should exclude all observations with missing data for variables that appear in any of the models.

Steve
sjsamuels@gmail.com



On Apr 3, 2011, at 6:48 PM, Richard Williams wrote:

"When contrasting models, just make sure they really are nested, e.g. you don't want to be imposing different proportionality constraints in your different models while at the same time dropping or adding variables."

At 04:15 PM 4/3/2011, Gao Liu wrote:
> Thanks, Nick. Right, it should be gologit2. So it is McFadden's Pseudo R^2.
> 
> Can I simply use the difference of the pseudo's R^2 of model with the
> focal variable and that without to explain the increased explanatory
> power? pseudo-R2 (with the variable)-pseudo-R2 (without the
> variable)=(L1(with)-L1(without))/L0. Can I use this result as the
> increased explanatory power? Are there any other ways to  answer the
> question of "how much more likely is the predictor able to predict the
> change of categorical dependent variable"
> 
> Thanks very much for your explanation.

First off, thanks to Nick for answering the first part of the Q. gologit2 uses the McFadden's Pseudo R^2, just like logit and ologit and assorted other Stata routines do. If somebody had a burning desire to calculate one of the other Pseudo R^2 measures it probably would not be too hard to do so.

With regards to the 2nd part of the question, I think the most customary thing to do is either a Wald test of the variable in question or a LR chi-square contrast between the model that has the variable and the model that doesn't. So, if the variable of interest was education, you could do something like

use "http://www.indiana.edu/~jslsoc/stata/spex_data/ordwarm2.dta";, clear
gologit2  warm yr89 male white age prst ed
test ed
nestreg, lr: gologit2 warm (yr89 male white age prst) (ed)

When contrasting models, just make sure they really are nested, e.g. you don't want to be imposing different proportionality constraints in your different models while at the same time dropping or adding variables.


-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   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