In linear regression, Cook's distance test is used to identify if there are
a few influential observations that affect the parameter estimates. I was
wondering if there is any such test to identify if a few observations
are unduly influencing the estimated results in the context of a
multinomial LOGIT model.
Thanks a lot! Please email to [EMAIL PROTECTED]
Sanjoy Ghose
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
