100
7 Let’s Start with Some Serious “Do Nots”!
1
10
100
1000
0
100
200
300
400
500
600
700
800
900
1000
(a)
(b)
Fig. 7.4 a Losses classes in M$ display a somewhat uniform width in log scale. b Once the same
classes are displayed in decimal scale the range compression phenomenon (for the lower classes)
becomes evident
Users generally avoid extreme values or statements if they can. For example, if a
score range goes from 1 to 5, many will select values in the 2–4 range as a result of
the “centering bias”.
Category-Definition Bias
Users oftentimes confuse frequency and probability. The confusion is not
solved by the risk matrix compilation guidelines generally offered to
users. The results are confused communications, as pointed out by Canadian CEOs (https://www.riskope.com/2016/03/23/canadian-ceos-are-looking-toimprove-measurement-risks-communication/). Researchers even talk about the “illusion of communication” generated by the “category-definition bias”.
7.8.2 “Newly Recognized” Risk Matrices Deficiencies
Ranking is Arbitrary
If the risk matrix uses an index approach (ranks from 1 to n), then indexes can be
in ascending or descending order. Both approaches are used in various industries.
Précédent

- 114/823

Suivant