This has been a very good discussion. I'm a bit fanatical on this because I've seen too many accidents where the facility performed a quantitative safety analysis, manipulated the result, assigned it a low priority, and did nothing. Let me add one more comment and then I'll shut up.

There's another reason why we don't try to plug numbers into a risk matrix (although I guess you could say that we're in effect using 1, 2 and 3). The greater the precision of a number, the more it's open to manipulation. It's easier to understand and challenge a 2 vs. a 3, than a 2.85 vs. a 2.87. And people make decisions all the time on numbers with 3 significant digits when the underlying data is precise to less than 1 significant digit.

As an example of manipulated numbers, here's a link to article on the January Brumadinho tailings dam collapse in Brazil that killed at least 272 people.

https://www.ecchr.eu/en/case/the-safety-business-tuev-sueds-role-in-the-brumadinho-dam-failure-in-brazil/

When the firm charged with certifying the dam's safety got numbers the company didn't like, they just changed the method and got different numbers. Whereas, in the qualitative risk matrix the consequence ranking would have been high, and the probability ranking would have been at least medium, given the known instability of the design and the fact that a similar dam - operated by the same company, no less - had failed a couple of years earlier.

Of course it's also easy to cheat on qualitative assessments. But I'd maintain that it's harder to get away with it.

Mike Wright

This is a very interesting topic that we are also struggling with as we are developing a risk management course for researchers with the Safety Training Consortium. We also settled on using the 3 by 3 grid rating likelihood versus severity of outcome of an incident and came to the conclusion that risk rating is very subjective. The problem is the unpredictability of incidents in research environments where experiments are constantly varied and their contexts may also change.

When thinking of risks in a research lab one should probably take a systems approach beginning with providing a safe research environment for the type of research to be conducted in a given space. At the end, there are the researchers who are conducting their experiments in the context of that environment plus other factors that may affect their experiments or their performance on a particular day.  Here, I think the risk assessment training comes in that empowers researchers to become rational judges of their experiment and the context where it is performed.  At this point, I would still stick to a simple risk rating grid as a visual guide to raise awareness to the fact that if a risk -after assessing what is involved - is medium to high for example, necessary minimization strategies need to be in place. I view the risk rating grid as a guide. Sandia offers a comprehensive risk assessment with percentages and some math that lends itself to a more standardized research setting.

My best,

Imke

UC Center for Laboratory Safety

While stumbling around the web with regard to thinking about the risk matrix, I came upon an article that questioned its value:

https://medium.com/**At_Symbol_Here**JornMineur/why-the-risk-matrix-must-die-620a7287e7c

The essence of the argument, I think, is that estimates of probability are very unreliable.   I'd appreciate the wisdom of the list regarding this essay and its conclusion.

Dave

David C. Finster
Professor Emeritus, Department of Chemistry
Wittenberg University

--- For more information about the DCHAS-L e-mail list, contact the Divisional membership chair at membership**At_Symbol_Here**dchas.org Follow us on Twitter **At_Symbol_Here**acsdchas

--- For more information about the DCHAS-L e-mail list, contact the Divisional membership chair at membership**At_Symbol_Here**dchas.org Follow us on Twitter **At_Symbol_Here**acsdchas