Criteria weights are typically cognitively demanding to elicit and numeric precision is problematic since information in real-life multi-criteria decision making often is imprecise. One important class of methods rank the criteria and receive a criteria ordering which can be handled in various ways by, e.g., converting the resulting ranking into numerical weights, so called surrogate weights. In this article, we analyse the relevance of these methods and to what extent some validation processes are strongly dependent on the simulation assumptions. We also suggest more robust methods as candidates for modelling and analysing multi-criteria decision problems of this kind.