We conduct an unlabeled DCE using 5 attributes each having 4 levels (complete enumeration method). Our expected sample size is 1000.
Given that our focus is estimated individual utilities, how would the number of versions we use effect the results? Should we rather use 10 versions instead of 100?
Thanks a lot for your support.