Re: st: orpobit vs ologit

[Richard Williams <Richard.A.Williams.5@ND.edu>]

At 03:08 PM 4/11/2007, Nick Winter wrote:
> Whatever the theoretical differences (in terms of assumptions about 
> the distribution of the latent response variable, conditional on 
> covariates; not the marginal distribution), in practice both will 
> give you substantively the same results.  (The coefficients will be 
> different because they are normallized differently, but the size of 
> effects in terms of substantively meaningful things like predicted 
> probabilities will be indistinguishable.)
I agree. The choice may just boil down to what the common practice is 
in your field. Seems to me like I see a lot of logit/ologit work in 
Sociology, and also in the occasional medical-related research that I 
read.  In economics I see probit/oprobit being used some.

The bigger issue may be whether you should use either ologit/oprobit 
in the first place.  The assumptions of the ologit/oprobit models are 
often violated.  Some programs try to deal with these violations, 
e.g. my own gologit2 and oglm programs available on SSC.  Another 
possibility is just to forego the ordinal route completely and 
estimate an mlogit model.  There are also other ordinal 
regression-related routines out there, like slogit and ocratio.

As noted before, Long & Freese's book is a good place to learn about 
these things.  See

http://www.stata.com/bookstore/regmodcdvs.html


-------------------------------------------
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/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/