Car accidents, cold, more examples of biased and misleading FMEA/PIGs results

Reviewing Car accidents, cold, more examples of biased and misleading FMEA/PIGs results adds to prior information published in this blogpost.

In particular we show how results of a Risk assessment matrix 5×5 or 4×4 can actually cloud our judgement.

Car accidents, cold, more examples of biased and misleading FMEA/PIGs results

Dear X,

Thank you for asking us to review your draft Risk Assessment for the ACME Inc. operation.
I reviewed your FMEA/PIG risk matrix 4×4 and have the following remarks. Note if it was a 5×5 matrix the remarks would be exactly the same.

How ready are you, really, to say that a “not likely”/”most unlikely” event with extreme consequences is IDENTICAL (from a risk assessment point of view) to an “expected”/”will occur” event with negligible consequences? (See the figure below, we are talking about the extreme cells of the “yellow” category).

Results of a Risk assessment matrix 5×5 or 4×4.

If you are really, really ready, or think it does not matter and we are “splitting hairs in four”, then consider that the first event is, for example, having a really critical car accident (with physical consequences for driver and passengers), whereas the second event is, for example again, catching a cold.

FMEA/PIG risk matrix (4×4): are the “not likely”/”most unlikely (1)” events with extreme consequences (4) IDENTICAL to “expected”/”will occur (4)” events with negligible consequences (1).  Risks are respectively 1*4=4 and 4*1=4….but, as shown in the text they are totally different!

If we judge by the number of safety devices in modern cars, roads, traffic safety campaigns etc.  I think you will concur that beside the factual cost of consequences and actual risk, we personally and societally understand/perceive these two types of events very differently. Their tolerance is completely opposite: one is considered intolerable (hence all the mitigations…), the other one, who cares (at best you wear a warmer coat, at worse you stay a couple days in bed…)! Yet, in FMEA/PIGs risk matrices 4×4 or 5×5 they come out with the same score (4 in the case at hand)…one more time demonstrating that these approaches are fallacious.

Examples of “not likely”/”most unlikely” event with extreme consequences that were/could have been “disregarded” by a conventional risk approach based on a matrix similar to the one you are proposing to use:
Fukushima, BP Gulf oil spill,  and finally, many mines’ tailings dams failures….
We could compile a very large list of these, of course, should we put some effort into this.

I look forward to discussing this further and hopefully to be able to deploy ORE (Optimum Risk Estimates) on your case to avoid your study to present biased and misleading results to Acme’s Management and to the public.

Tagged with: BP Gulf oil spill, car accident, cold, FMEA/PIG, Fukushima, Optimum Risk Estimates, Tailings Dams, tolerance

Category: Consequences, Optimum Risk Estimates, Probabilities, Probability Impact Graphs, Risk analysis, Risk management, Tolerance/Acceptability