When modeling choice behavior, the ranking patterns were corresponding to the multiple paired comparison.
I'm wondering:
1. Whether these two psychological process are really the same?
2. The number of patterns is as large as 2m where m is 2-combination of the number of options. Therefore, some patterns may has few persons and the estimation will be inaccurate.
3. Maybe we can reduce the number of patters through the recursive property of pairwise comparison. For example, if j>k, and k>l, then (j,l) pattern may be remove from the pairwise outcomes.