Thread Subject:

mixed effects logistic regression

Hi,

I'm trying to run a mixed effects logisitic regression in Matlab with binary responses and a matrix of 8 regressors, 13,000 total obs, and 90 even sized groups. If I am thinking about this correctly, I don't need to specifiy a nonlinear function, but just need to change the observation error from normal to logistic. In this case, does anyone know how to change the dist'n of the observation error term, e_ij? If this is not the case, can someone point me in the right direction of how to call nlme if I want to run this mixed effects logit regression with the specified structure above? Just to be sure, my model is a linear combination of the 8 regressors, which is why I thought maybe I don't need to specify a nonlinear form, unless I were to explicitly model it as a logistic function?? Much thanks in advance for any help.
CF