# 158
Dear Bartosz Paradowski, B.Paradowski, Bartłomiej Kizielewicz , Wojciech Sałabun
I read your paper
Equal Criteria Influence Approach (ECIA): Balancing Criteria Impact in Multi-Criteria Decision Analysis
My comments
1- In your abstract you say “Traditional approaches to weighting criteria often result in insufficient consideration of the varying influence of criteria on decision outcomes”
Not in my opinion; weights are only useful to gauge the relative importance between criteria. They do not have the capacity to evaluate alternatives. If you wish we can discuss this in extent either in RG or by email. Mine is [email protected], my phone +1 613 770 7123 (Canada) About 6 hours behind yours
“ECIA emphasizes the impact of criteria on the preference of alternatives”
Again, weights do not have any influence on alternatives. If they had there would be as many evaluations of the same problem, as the number of intervening. MS Who of them is right?
Criteria are lineal equations, and as that, they can be represented in a graphic by straight lines. If you multiply the values of a criterion by a weight, then the line will displace parallel to itselfaccording to the weight, since all its values are multiplied by the same quantity. In so doing, they certainly will have a different value for each alternative, but this is not an evaluation, it is only a consequence of a geometric representation.
According to Shannon, the evaluation from a criterion is given by the way in which its values are dispersed, measured by entropy. Consequently, the impact that you mention affect to all performance values within a criterion, in the same manner. Different would be if you multiply each of these values for a different weight, because in this way you may alter the dispersion or discrimination of the criterion.
2- In page 2 the paper makes a simplistic explanation between subjective and objective weights, assuming that both are adequate for evaluating alternatives. This is not so. Weights are trade-offs that Saaty assumed (his words) as weights (without any explanation), when in reality they are two different things, and with different purposes.
3 - I grant that intuitively it seems that weights are useful to evaluate alternatives, not in my opinion, as I explained above. Weights are derived from the values given by Eigen Value Analysis and normalized to 1, which gives way to trade-offs, and used in compensatory methods. However, they are also being applied to MCDM methods that are not compensatory, as PROMETHEE, ELECTRE, and many others. Do you think correct to use trade-offs as weights?
Further, and with reason, your paper says that it is important the contribution of the DM, and I plenty agree. What should be done, is to multiply the weights by each performance values in each criterion, and after that, use entropy to evaluate criteria. This will displace unevenly the existing distances between criteria, and from then on, entropy gives the importance of each criterion to evaluate alternatives. Much better that this is not using weights, run the software, and then, examining the results, the DM using his experience, know-how, research, exogeneous variables, etc., can change whatever he wants, but not by intuition but on analysis.
4- In page 3 “This approach is based on MCDA sensitivity analysis; wherein individual criteria are systematically excluded from the decision matrix”
This is procedure has been rejected by most researchers. It does not make sense to change only one criterion and keeping the others constant. The reason, is first, you do not know which is the most important criterion (its weight has noting to do with it, it is only intuitive, with not mathematical support), and secondly, a MCDM problem is a system, and as such, a change in one criterion of component of the system, may affect the others. It is like decreasing the out put of a car engine, without considering how it affects the speed, acceleration, the losses in the car body, its transmission, etc. Of course, the paper enumerates a lot of advantages,using only words, without any proof of what it asserts.
These are some of my comments. I hope they can help
Nolberto Munier