The Average Does Not Matter
Biologist Stephen Jay Gould was told he was dying of a rare form of cancer. His time on this earth was limited. Depression set in. However, a strange thing happened on his way to the Grim Reaper. His research found that people with his characteristics were in the statistical tail of the mathematical probabilities related to the chance of death. Instead, there was a very good chance that he would beat the normal odds of survival. Indeed he did, and his article, “The Median Isn't the Message,” became a classic of hope for anyone facing what appears to be long odds.
The 80-20 Rule
Economist Vilfredo Pareto had discovered something similar nearly a century earlier. He found that it wasn't the average that mattered but rather the distribution. Few things truly follow a bell curve. Understanding where you are relative to the fat side or the skinny side of the distribution was essential for understanding and mitigating problems. Pareto became synonymous with what we now term the “80-20” rule. His Pareto Diagram is today widely considered the first tool of reliability analysis.
Life is Skewed
Hydrologist H.E. Hurst made a similar conclusion in the first half of the twentieth century as he discovered the secret to building safer dams on the Nile River. Civil engineers still use his techniques today. In the second half of the twentieth century, mathematician Benoit Mandelbrot used the foundational concepts of skewness and asymmetry to develop fractal geometry. Mandelbrot's book, "The (mis)Behavior of Markets," remains a classic for understanding financial risks in the stock market.
Even in Work Order Management
My colleague Jim Oldach re-taught me the power of the Pareto Diagram several years ago when it came to Work Order Analysis. The mean or median matters very little in terms of the quantities of different types of work orders. If we do the Pareto Analysis, we’ll usually find a 90-10 or even a 95-5 relationship – in other words, we are spending 90 or 95 percent of our time fixing the same handful of problem types. Focus on these and most of our reliability issues will be resolved.
Complexity Requires A Different Approach
So why have so many technical analysts defaulted to simple statistics like the mean? We are taught to lean too heavily on breaking complex systems into more simple modules. In turn, simple statistics are easier to apply to less complex modules. The trouble is that simplifying and applying "normal" statistics requires us to assume that events are independent and tend to migrate to a central tendency.
In the real world, most things are richly skewed and dependent. This is why most truly worthwhile things in life – your spirituality, your family, your friendships, and your health – are not simple. The median, and especially not the mean, is not the message.
Communicate What Matters Most
Provide some insight into where and when the big stuff matters most. And in some cases, especially when the odds are long, provide some insight into when there may be power in the small stuff.
Have endless respect for your decision maker and their ability to truly understand. Yes, it takes more analysis time and communication skill on your part. It is, however, your duty as a trusted advisor. It is one of the keys to arriving at quality decisions.
The illustration “Don’t Sweat the Small Stuff” and excerpts from this article are taken from my book, “Communicating Reliability, Risk and Resiliency to Decision Makers: How to Get Your Boss’s Boss to Understand.” The second edition will be released in June 2022. Sign-up for updates at Communicating with FINESSE.
Comments