Re: st: Stata's pseudo R-squared for oprobit

[May Boggess <mboggess@stata.com>]

On Wednesday afternoon Ramani asked about the pseudo-R^2 reported by 
-oprobit-:


> I have not been able to find a definition of the particular pseudo
> R-squared that Stata reports with the results of an oprobit regression
> in either the manuals or in the list archives.
> As I understand, there are many pseudo R-squareds around. Can someone
> please tell me whether the one Stata reports is Maddala's (1983) or
> Zavoina and McKelvey's (1975) or something else?
The formula used is used with many Stata estimation commands.
The pseudo-R^2 is defined as 1 - (LL/L0) where LL is the maximized log
likelihood and L0 is the maximized log likelihood of the constant only 
model. Here is an example:

 clear
 sysuse auto
 oprobit rep78 mpg len
 display "pseudo R^2 = " 1-  e(ll)/ e(ll_0)

Ramani can find this formula in the Reference manual [R] maximize, in 
the Methods and Formulas section (top of page 144, in the version 9 manual).

 -- May
mmb@stata.com
*
*   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/