Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
omitted variables in logistic regression
0 views
Skip to first unread message
Peter A. Lachenbruch
Jul 27, 1993, 7:01:00 PM7/27/93
to
At convergence, the coefficients are the solutions to equations of the form
(X'WX)**-1X'WY, so one might consider looking at the transformed variables
Z=W**(1/2)Y and U=W**(1/2)X and carrying out the same analysis that you did
for linear regression. I believe the analogy carries through exactly.
Tony Lachenbruch
(X'WX)**-1X'WY, so one might consider looking at the transformed variables
Z=W**(1/2)Y and U=W**(1/2)X and carrying out the same analysis that you did
for linear regression. I believe the analogy carries through exactly.
Tony Lachenbruch
0 new messages