|
Conclusion
All aspects of risk analysis, from planning to
the interpretation of results, rely on subjectivity.
This article shows that in the third stage of
the process, hazard analysis, the derivation of
values for both consequence and probability is
based on numerous judgmental decisions. Moreover,
the method most used for determining cause and
effect and for combining probabilities, fault
tree analysis, was seen to be dependent on human
factors for both its construction and its use.
While the language of the article suggests that
quantitative analysis is being addressed, it should
be added that qualitative analysis, extensively
used on many of today's systems, is by definition
based on human judgment.
It was seen that inputs to FTA may arrive from
disparate sources, of varying pedigree and trustworthiness,
so the error in the estimation of probabilities
can be great without there being any human sensitivity
to it. It would therefore be useful to introduce
an activity in risk analysis in which information
and its source and pedigree were reviewed. There
should then also be a formal requirement to place
a confidence level on information and to make
a decision on whether there is a need to seek
further information from a source of higher pedigree.
It would also be useful to revise the risk-analysis
syllabus to cover the ways in which subjectivity
is introduced and the effects that it has, and
to make the process's assumptions more explicit
to analysts. Thus, analysts would be taught to
understand not only mathematical assumptions but
also their own human biases. There would then
be an increased chance that subjectivity would
be considered, and partly neutralized, during
the analysis and management of risk, and that
evidence would exist for placing confidence figures
on results. For an introduction to human biases,
see the author's earlier article [Ref. 8].
While this article points to the subjectivity
in hazard analysis and invites analysts to understand
it better, it should also be emphasized that human
creativity and decision-making are major strengths
of the process. Transparency and a methodical
approach are others. In spite of the criticism
that it receives, risk analysis and its techniques
serve industry well and could beneficially be
transferred from science and engineering to other
domains where there is a recognized need for a
better understanding of risks.
About the Author
Felix Redmill is a consultant in risk management,
project management and quality improvement, and
the author and/or editor of many articles and
books on risk and other technical subjects. He
spent more than 20 years in industry, as engineer
and manager, in both communications and system
development, before starting his consulting business.
Felix has been the Coordinator of the Safety-critical
Systems Club (U.K.) since its inauguration in
1991. He has been a visiting lecturer at various
universities, served as chairman and member of
various professional committees, and been a long-standing
member of the European Workshop on Industrial
Computer Systems. He was the Awards Banquet speaker
at the 18th International System Safety Conference
in Orlando, Florida.
References
1. Redmill, F. "Risk
Analysis - A Subjective Process." In
eJSS, Vol. 1, No. 1, February 2003.
2. Engineering Safety Management Guidance
(The Yellow Book). Railtrack, issue 3, on
behalf of the U.K. Rail Industry, 2000.
3. Development Guidelines for Vehicle Based
Software. The Motor Industry Software Reliability
Association, U.K., 1994.
4. File On Four. BBC Radio 4, December
12, 2000.
5. Fischhoff, B., P. Slovic and S. Lichtenstein.
"Fault Trees: Sensitivity of Estimated Failure
Probabilities to Problem Representation."
In Journal of Experimental Psychology: Human
Perception and Performance. Vol. 4, No. 2,
pp. 330-344, 1978.
6. Funtowicz, S.O. and J.R. Ravetz. Uncertainty
and Quality in Science for Policy. Dordrech:
Kluwer, 1990.
7. Freudenburg, W.R. "Heuristics, Biases,
and the Not-So-General Publics: Expertise and
Error in the Assessment of Risks." In Social
Theories of Risk. S. Krimsky and D. Golding,
eds. Westport: Praeger, 1992.
8.
Redmill, F. "Some Dimensions of Risk Not
Often Considered by Engineers." In Journal
of System Safety, Vol. 38, No. 4, Fourth Quarter
2002.
Acknowledgment:
An earlier version of this article was published
by the IEE.
|
