Is there anyway to get something resembling a U² statistic using aggregate logit? The U² statistic was explained to me as the "percent uncertainty explained“.
The model is based on a choice task with 2 alternatives across 8 screens where the response, instead of being a discrete choice, is a scale 1-5 where 1=very likely to buy product on the left, 5= very likely to buy the product on the right, 3=indifferent. Choice data are then modeled like a chip allocation where scales are assigned probabilities. i.e. an answer of 1 gets probability of 1 for the left and 0 for the right, etc. (.75/.25, .5/.5, .25/.75, 0/1 for responses 2,3,4,5).