Page 1 of 1

constant term for ordered logit

Posted: Tue Mar 01, 2011 11:55 am
by s2501
hi, i understand that in the eviews guide, it is stated that "A separate constant term is not separately identified from the limit points gamma, so EViews will ignore any constant term in your specification."
however, is it possible to obtain the constant term from the output?
also, what does Limit Points represent? slightly confused here.

thanks!

Re: constant term for ordered logit

Posted: Tue Mar 01, 2011 1:22 pm
by EViews Glenn
There really is no single constant term per se in an ordered logit. There are a "set of constant points" which define the probabilities for each choice for an observation having the value of 0 for all of the remaining explanatory variables. This is the reason that EViews ignores any "C" in your specification; it is estimating a set (one less than the number of categories) of these values.