Multi-criteria decision aid (MCDA) methods have been around for quite some time. However, the elicitation of preference information in MCDA processes, and in particular the lack of practical means supporting it, is still a significant problem in real-life applications of MCDA. There is obviously a need for methods that neither require formal decision analysis knowledge, nor are too cognitively demanding by forcing people to express unrealistic precision or to state more than they are able to. We suggest a method, the CAR method, which is more accessible than our earlier approaches in the field while trying to balance between the need for simplicity and the requirement of accuracy. CAR takes primarily ordinal knowledge into account, but, still recognizing that there is sometimes a quite substantial information loss involved in ordinality, we have conservatively extended a pure ordinal scale approach with the possibility to supply more information. Thus, the main idea here is not to suggest a method or tool with a very large or complex expressibility, but rather to investigate one that should be sufficient in most situations, and in particular better, at least in some respects, than some hitherto popular ones from the SMART family as well as AHP, which we demonstrate in a set of simulation studies as well as a large end-user study.
展开▼
