Also is linear regression acceptable to bayesians since no normality required.++My 
understanding is that logit first fits a linear regression on the dichotomous Y and 
then forces this to lie between 0 and 1 via the logit y=1(1+e^x), finally the log 
likelihood is calculated as the P(estimate|D) - was this P(estimate|d), I forget the 
commonly cited log-likelihood for logit but was it calculated via bayes 
theorem?++Sorry about newbie question..


-------------- ���ͳ� ī������ KORNET -------------

.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to