AHP has a lot of drawbacks such as most of the MCDM. Decision making process is uncertain. AHP like the other MCDM methods tries to tackle the uncertainties while the decision makers have the right to decide in the light of the output of these MCDM processes.

I am an AHP advocate, however I admit all the associated drawbacks.

Dear Mohamed

Thank you for answering my question.

It is true that decision-making is per se uncertain, but some methods such as AHP and ANP add more uncertainty, by working with intuitions and with pair-wise comparisons and assigning arbitrary values to that preference.

Of course, the fact that another DM may think differently and generate a different solution appears not to bother the stakeholders.

Since you accepted my request, perhaps you can illustrate to me the logic in telling for instance that comparing two football teams, you establish that criterion quality is 3 times more important than the criterion speed.

Of course, everybody has the right to have preferences between two criteria and say that prefer one over the other, nothing wrong with this, but give a quantitative value for that preference? Can you say that tenderness is 2 times more important than love? Is this lack of rationality that makes AHP perhaps an interesting concept from the conceptual point of view, but a complete non-sense in real-life projects.

Is this one of the drawbacks that you admit in AHP?

And there are many drawbacks like this.

However, you are absolutely right in saying that decision-makers have the right to decide in the light of the output of these MCDM processes. This is the bottom-up approach, that in my opinion, is much better than the top-down procedure, where the DM alters at will data with unproved values.

However, you contradict yourself, since this bottom-up procedure is against AHP philosophy, where the decision matrix, composed of invented values, is something that does not happen with most if not all of the MCDM methods, which also have the advantage to use entropic and the objective weights.

You can certainly apply AHP to trivial problems which result affects only you, but you can't use AHP in large problems that affect the lives of thousands if not millions of people, because, as AHP does, the DM decides for them.

This is not theory. The construction of the Itaipu Dam in Brazil displaced thousands of people from their lands and houses and many had to immigrate to another country. What about the construction of the Three Gorges Dam in China, where millions were displaced?

Who authorized the DM to be the voice of those people? Would you like that your property, the course of your life, and perhaps your work, to be altered because a DM or DMs decides so? Have you thought about this?

Arrow, in his very well-known theorem, establishes that it is a dictatorship, and thus, not admissible.

You have the right to advocate whatever MCDM you want, and nobody has the right to challenge it

Hi every one!

for this method I noticed in some cases, when we give high appreciation for one criterion or more against another less appreciation (for some other criteria), the method is unable to fulfill the non-dictatorship property or at least to alert a non-specialist user (case of my student) about that.

Indeed, some choices (from Saaty table for criteria scales ) can make easly dictator criteria, in our case, after calculation, the method indicated a very good coefficient of consistency which promise for a very good weights. So how can avoid this situation!

For numerical results, please contact me !

Dr. Zhor CHERGUI

Dear Zhor

When you mention non-dictatorship I guess that you are referring to the Arrow Impossibility Theorem, which establishes that taken decisions by others, as in AHP, is a dictatorship, a concept with which I am in complete agreement and have mentioned many times in RG.

I don't judge about the goodness of the Saaty fundamental ratio scales, however, I am against the fact that the DM is FORCED BY A FORMULA, to modify his appreciations.

This "analysis", based on transitivity may be good to gauge how consistent the DM is with his appreciation. But, please tell me, because I don't see it, what relation does it have with the real world that is mostly INTRANSITIVE, and on WHAT GROUNDS this is applied to the real world. In other words, why is it assumed that their results, based on intuitions of the DM, can be replicated in the real world? Is there any theorem or axiom about that?

Over the years in RG, I have formulated several times these two questions, and curiously, I never got somebody supporting with logical arguments this aspect of AHP.

It appears that no explanation is possible, and then, the whole method is based on false premises. I am not against the AHP method, what I am against to is its lack of realism, by forcing reality to adjust to theory.

I would very much appreciate it if you can send me the calculations you mention, for me to have a look at them.

Nice discussion thread, thanks for proposing it. The main problem seems to be with the subjective choice of criterion weights, but isn’t this a problem with MCDM in general? As I see it, the weights for different criteria should be proposed by different stakeholders as part of a participatory process, the AHP pairwise comparison then allows different viewpoints to be compared. Effectively, we would then have different outcomes depending on the importance (power?) of different stakeholder communities. Thus I would see AHP as simply a tool for analysing and evaluating different policy options together with decision making stakeholders. In this sense the subjectivity of the process is a strength - each decision centre has its own perspective and thus logically its own subjectively determined weighting. This clarifies potential conflicts and power relations and allows marginalised stakeholder opinion to be included in a non-confrontational way. The use of arbitrary weights to produce a supposed optimum solution is a problem of poor research design, not a problem with AHP per se.

An issue can occur when you remove or add an alternative. In general, the remaining ranks are not maintained (pairwise comparison). The well-known issue of rank reversals.

Dear Marco

True, but it happens with most MCDM methods, although it was discovered in 1983 in AHP

Richard Hewitt

Thank you, Richard, for your answer

RH-Nice discussion thread, thanks for proposing it. The main problem seems to be with the subjective choice of criterion weights, but isn’t this a problem with MCDM in general?

NM- Yes, in general, it is, but there are MCDM methods that don’t use them.

If you want to go use weights, you can use objective weights, from entropy and using CRITIS or another statistical method. Nevertheless, if you use objective weights they can be used for evaluation alternatives because they are the quantity of information contained in each criterion. Subjective weights may be used to evaluate criteria but not alternatives, because they are no weights but trade-offs, good for performing a balance, but not for evaluation alternatives.

RH-As I see it, the weights for different criteria should be proposed by different stakeholders as part of a participatory process, the AHP pairwise comparison then allows different viewpoints to be compared.

NM- Agreed, and then you must have a procedure to consolidate the different weights.

RH - Effectively, we would then have different outcomes depending on the importance (power?) of different stakeholder communities.

NM- Absolutely correct

RH-Thus I would see AHP as simply a tool for analysing and evaluating different policy options together with decision making stakeholders.

NM- Again, weights can’t be used to evaluate policy options. You don’t need AHP for that. You must consider all the different options. All options and values from stakeholders must be considered. Let the software take care of the significance of each one. In this way, however, the DM is, the result will be always the same.

RH - In this sense, the subjectivity of the process is a strength - each decision centre has its own perspective and thus logically its own subjectively determined weighting.

NM – I like when you say ‘each decision centre has its own perspective’. It is true, however, can each centre determine its own weight, based on what and compared to what? In addition, it is not how AHP operates.

RH - This clarifies potential conflicts and power relations and allows marginalised stakeholder opinion to be included in a non-confrontational way. The use of arbitrary weights to produce a supposed optimum solution is a problem of poor research design, not a problem with AHP per se.

NM – could you explain to me how this process may clarify potential conflicts? There is a mediator with enough authority to do that? In real-world it does not happen

RH – ‘The use of arbitrary weights to produce a supposed optimum solution is a problem of poor research design, not a problem with AHP per se’

NM- You lost me. How this discrepancy can produce an optimum result? Could you explain under what mechanism?

If for instance, the design is poor, usually there is somebody called ‘Project Manager’ who can judge it, not AHP of course.

.

Suggest reviewing this paper for MCDM methods:

https://www.sciencedirect.com/science/article/abs/pii/S092583881732827X