Discrete choice models question

Hello, all,

I am a cognitive scientist trying to model how people select various
items on a web page. My question refers to whether a multinomial logit
model is appropriate for these kind of data or perhaps a different
model may be needed.

This is my problem: when a person sees a new web page, they decide
which link to select to click next based on several variables
(relevance to their search goal, position in page etc.). I would like
to predict where the person would click next (ie., which link they
choose), based on the value of these variables. However, the problem
is that there is not a fixed set of alternatives for any decision point
(that is, if you click on a link, you'll get a new set of choices).
Would multinomial logit work for this setting?

Any thoughts about this would be greatly appreciated.
Thank you in advance,
Raluca Budiu

.