EViews.com

Dear all,

As part of my economics dissertation i am carrying out an experiment to assess the impact of time delays in games. I need to device an unordered logit or unordered probit model. Can anyone suggest how i can do this with eviews6? My dependent variable takes values from 1-5, where each number represents an observed pattern of behaviour. My other variables include general demographic stats ex, age, gender, occupation etc.

Can anyone help on this

Brian

Unfortunately there is nothing built-in for that model. There is, however, a logl object example that estimates the multinomial logit. It will need a bit of modification for your particular data structure, but it shouldn't be too bad.

See Help/Quick Help Reference/Sample Programs & Data and follow the logl link.

I'm not sure (Glenn: can u comment on this), you can use a count model for such dependent variable.

Count models aren't unordered.

Startz is most certainly correct. Moreover, count models are both ordered and cardinal.