Page 215 - Materials Chemistry, Second Edition

10.5 Challenges in the application of MADM for LCSA 213
10.5.1 Choice of MADM method
A detailed review by Greco et al. (2016) shows that many methods (more than 100) exist for
solving discrete decision problems. Choosing an MADM method in itself can be posed as an
MADM problem, as suggested by Guitouni and Martel (1998). The first step in using the
MADM approach, as explained in Munda (2005), is to take a stand (value choice) whether
to adopt compensatory approach or noncompensatory approach. Choosing any of these
approaches automatically results in following “weak” sustainability assessments (in the case
of MADM methods based on compensation principle) or strong sustainability assessments
(in the case of noncompensatory assessments) (Kalbar et al., 2017a; Rowley et al., 2012;
Munda, 2005). Kalbar et al. (2017a) specifically demonstrate that linear weighted sum
(LWS), a more straightforward form of compensatory approach, favors extreme solutions.
Each of the MADM methods uses different mathematical principles, and hence, it is nec-
essary to test more than one MADM methods in LCSA. For example, Kalbar et al. (2017a)
report that the use of distance-based MADM (TOPSIS) is a more suited approach than a
relative utility-based approach such as linear weighted sum (LWS) when ranking the individ-
ual’s environmental footprint.

10.5.2 Rank reversal in MADM
Rank reversal, meaning change in ranks of the alternatives due to change in the MADM
methods, or addition, or deletion of criteria, or change of weights, is a well-known and
well-discussed phenomenon. Almost all methods of MADM has the rank reversal property
(Mousavi-Nasab and Sotoudeh-Anvari, 2018; Mufazzal and Muzakkir, 2018). It can be con-
cluded from the studies dealing with rank reversal that rank reversal is unavoidable and
is an underlying property of the MADM approach. However, in some cases (e.g., problems
with dominating alternatives) application of different MADM methods can result in selecting
the same alternative as the most preferred one.
One of the ways to handle rank reversal is restructuring the decision making, e.g., scenario-
based decision making, as demonstrated in Kalbar et al. (2012). The basic approach in
scenario-based decision making is defined as the case/situation-specific weights. By apply-
ing the case-specific weights, more consistent ranking will be generated by any of the MADM
methods.
Hence, it is recommended to structure the decision-problem correctly by articulating
scenarios, and more than one method can be used to identify the most preferred alternative.
Spearman’s rank coefficient can be used for checking the agreement between the ranks
generated by two different MADM methods and if there are more than two MADM
methods used, Kendall’s coefficient of concordance can be used, as demonstrated in Kalbar
et al. (2015).

10.5.3 Dominating alternatives

In a decision problem, there could be multiple numbers of alternatives. Alternatives can be
divided into two groups, such as alternatives that are dominated and nondominated alterna-
tives (Kalbar et al., 2017a). An alternative can be called dominated if there exists another