Re: Interpreting the coefficient in a GLS binomial model

If I understand correctly, you're assuming the cumulative logistic distribution function for an index function model.
If you're looking for the marginal effect on the odds ratio rather than on the probability, you want:

p = e(x'B)/(1+e(x'B))

This follows because your odds ratio is: p/(1-p) = e(x'B)

and your log odds ratio: ln p/(1-p) = (x'B)

Assuming your regressor is your ideological index, then, simplifying a bit:
p= 1/(1+exp(2.982868 + (-1.450829*0.5)))

I hope this is what you're after.
.