If you are talking about measurement in fuzzy linguistic scales, this measurement is robust, i.e. the natural small errors that may occur in constructing the membership functions do not have a significant influence on the selection of the optimum set of values.