Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


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

Re: st: ologit, gologit2, mlogit?


From   Richard Williams <[email protected]>
To   [email protected], [email protected]
Subject   Re: st: ologit, gologit2, mlogit?
Date   Mon, 08 Jul 2013 12:56:36 -0500

mlogit may be your best bet (although there are a few other choices, such as slogit). But, just from what you say, I wonder if your model is too complicated and/or you are spreading your data too thin. You might consider combining categories of the ordinal variable (at least if some have very small frequencies), or using fewer variables, or (if using autofit with gologit2) use the .01 level of significance. gologit2 troubleshooting tips are at

http://www3.nd.edu/~rwilliam/gologit2/tsfaq.html

You might also want to assess how substantively important the violations are -- you give up a lot of parsimony when you switch from ologit to mlogit, and it may not be worth it if the violations are substantively trivial. You might consider using BIC tests as an alternative criterion for rejecting the proportional odds model, e.g. do something like

gologit2 y x1 x2 x3, pl sto(ologit)
gologit2 y x1 x2 x3, npl sto(gologit)
lrtest ologit gologit, stats

The likelihood ratio test is similar to the Brant test. But, you'll also get BIC stats reported, and it is possible that the BICs might favor the ologit model even if the LR test says to reject it.

At 03:57 AM 7/8/2013, John Antonakis wrote:
Hi:

I have model, having a dependent variable with 5 categories that are ordered. The model violates the assumptions of the -brant- test (after ologit estimation). It also produces predicted probabilities of less than zero in the user-written command (from SSC) -gologit2- (generalized ordered logit); in addition, I can only estimate a partial model because I can't estimate the full model with some dummy controls.

I think that the only option I have available is to estimate a -mlogit- model, which makes no assumptions on the ordering or on the proporational odds assumption (brant).

Does anyone know of any literature to support my intuition on this point?

Or are there any other ways to estimate this model with Stata?

Thanks!
John.

--
__________________________________________

John Antonakis
Professor of Organizational Behavior
Director, Ph.D. Program in Management

Faculty of Business and Economics
University of Lausanne
Internef #618
CH-1015 Lausanne-Dorigny
Switzerland
Tel ++41 (0)21 692-3438
Fax ++41 (0)21 692-3305
http://www.hec.unil.ch/people/jantonakis

Associate Editor
The Leadership Quarterly
__________________________________________

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

-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  [email protected]
WWW:    http://www.nd.edu/~rwilliam

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


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index