Kruskal's gamma as I know it would require
you to bin the predicted probabilities
before you -tabulate-.
Rather more information is shown by
a plot of observed response versus
predicted response. This is always
possible and sensible regardless of
the number of predictors.
A canned way of doing this is -ovfplot-
in the -modeldiag- package.
That the correlation between observed
and predicted is not crazy follows from
arguments in
Zheng, B. and A. Agresti. 2000. Summarizing the predictive power of a
generalized linear model. Statistics in Medicine 19: 1771-1781.
(I believe that Professor Agresti is in your Dept.)
Nick
[email protected]
Dinesh Sharma
> I am running a logistic regression with large data set (n=50000) and
> need to calculate the Kruskal's Gamma coefficient between the
> dependent
> variable y and the predicted probabilities. Is there any
> stata routines
> available for this? I tired tabulate command, which obviously did not
> work. Any suggestion is highly appreciated.
*
* 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/
