| |
|
Science Forum Index » Statistics - Math Forum » Pseudo R2
Page 1 of 1
|
| Author |
Message |
| Timothy Mak |
 Posted: Tue Jan 02, 2007 6:18 am |
|
|
|
Guest
|
Hi,
Consider the usual formulation for a logistic regression model:
Logit p = b0 + b1X1 + b2X2 + ...
Does anyone know of, or can calculate an approximation for the
Expected(Pseudo R2), given X is multivariate normal and b fixed?
For the unfamiliar, the formula for McFadden's Pseudo R2 is:
1 - Loglik(model)/Loglik(null model)
where null model includes only the intercept.
If it helps, you can assume b0 is 0, and X has mean 0 and variance 1
and correlation rho between all pairs.
Thanks very much,
Tim |
|
|
| Back to top |
|
| |
|
Page 1 of 1
All times are GMT - 5 Hours
The time now is Fri Oct 10, 2008 8:22 pm
|
|
