With criterion ratings, sensitivity testing arises from errors that may derive from
the actual estimation of the valuations themselves. They may be based on incomplete
data or, in the case of qualitative assessments, may be derived from subjective
judgements by specialists from the relevant field. Their exact value may therefore
not be known with complete confidence.
The process can be arduous and time-consuming, involving a large number of
iterations. As different criterion and weighting scores are varied incrementally, the
effect of these changes on the original ranking of options is gauged. Incremental
changes in criterion valuations possessing a high weighting score are more likely to
alter the baseline ranking than changes in the scores of less significant criteria.
Particular attention may also be paid to criterion scores that involve a high degree
of uncertainty and subjectivity. A competent decision-maker should identify such
criteria and analyse the effect of their variation on the overall ranking.
Example 15.2 Using a Sensitivity Analysis Within the SAW Method
A timber manufacturing firm wishes to open a new plant and six alternative locations
have been identified. The sites, all in different countries, are evaluated using the
following five criteria:
1. C 1 – Cost of land purchase (Rs. million)
2. C 2 – Building costs (Rs. million)
3. C 3 – Labour costs (Rs. thousands)
4. C 4 – Ease of transportation (10-point scale)
5. C 5 – Suitability of climate (10-point scale)
The performance matrix for the six sites is indicated in Table 15.4.
(i) Rank the six options using the simple additive weighting model
(ii) Gauge the effect that a Æ 50% variation in weightings for the first three criteria
(C 1 , C 2 and C 3 ) will have on the baseline rankings
Solution 15.2 Part (i)
Because three of the criteria have zero valuations on one of the options, the
transformation process employed in the previous example cannot be used for these
Table 15.4 Raw criterion scores
Criteria of evaluation
Option
Land
Building
Labour
Transport
Climate
(Rs. million)
(Rs. million)
(Rs. thousand)
(0–10)
(0–10)
Site 1
0.00
1.20
12.50
2.50
2.50
Site 2
2.00
12.00
72.00
5.00
5.00
Site 3
2.00
5.50
5.00
5.00
5.00
Site 4
2.00
3.00
5.00
5.00
5.00
Site 5
17.90
0.00
3.80
10.00
7.50
Site 6
0.60
10.90
0.00
7.50
10.00
Weight
0.28
0.28
0.25
0.15
0.04
15.4 Sensitivity Testing
419
the actual estimation of the valuations themselves. They may be based on incomplete
data or, in the case of qualitative assessments, may be derived from subjective
judgements by specialists from the relevant field. Their exact value may therefore
not be known with complete confidence.
The process can be arduous and time-consuming, involving a large number of
iterations. As different criterion and weighting scores are varied incrementally, the
effect of these changes on the original ranking of options is gauged. Incremental
changes in criterion valuations possessing a high weighting score are more likely to
alter the baseline ranking than changes in the scores of less significant criteria.
Particular attention may also be paid to criterion scores that involve a high degree
of uncertainty and subjectivity. A competent decision-maker should identify such
criteria and analyse the effect of their variation on the overall ranking.
Example 15.2 Using a Sensitivity Analysis Within the SAW Method
A timber manufacturing firm wishes to open a new plant and six alternative locations
have been identified. The sites, all in different countries, are evaluated using the
following five criteria:
1. C 1 – Cost of land purchase (Rs. million)
2. C 2 – Building costs (Rs. million)
3. C 3 – Labour costs (Rs. thousands)
4. C 4 – Ease of transportation (10-point scale)
5. C 5 – Suitability of climate (10-point scale)
The performance matrix for the six sites is indicated in Table 15.4.
(i) Rank the six options using the simple additive weighting model
(ii) Gauge the effect that a Æ 50% variation in weightings for the first three criteria
(C 1 , C 2 and C 3 ) will have on the baseline rankings
Solution 15.2 Part (i)
Because three of the criteria have zero valuations on one of the options, the
transformation process employed in the previous example cannot be used for these
Table 15.4 Raw criterion scores
Criteria of evaluation
Option
Land
Building
Labour
Transport
Climate
(Rs. million)
(Rs. million)
(Rs. thousand)
(0–10)
(0–10)
Site 1
0.00
1.20
12.50
2.50
2.50
Site 2
2.00
12.00
72.00
5.00
5.00
Site 3
2.00
5.50
5.00
5.00
5.00
Site 4
2.00
3.00
5.00
5.00
5.00
Site 5
17.90
0.00
3.80
10.00
7.50
Site 6
0.60
10.90
0.00
7.50
10.00
Weight
0.28
0.28
0.25
0.15
0.04
15.4 Sensitivity Testing
419