QUESTION: Can Rare Events Be Meaningfully Measured?February 12, 2013 by Stacey Barr
Subscriber Nia B. asks: “We need to establish a measure for safety, however we have a terrific safety record and insufficient data to establish one. Is there a meaningful way to measure this?”
Most measures of performance are used to track areas that need improvement. But there is a case to measure things that are performing well. And areas like innovation, industrial disputes, hospital infections and safety performance are good examples: any movement in the wrong direction is something you want to know about, even if it’s not something you’re actively trying to improve.
But when things like safety are going well, they’re not producing any data to monitor! You might get one accident every year or so. And that makes for a stupid-looking and very sparse bar chart.
So you end up resorting to next-to-useless measures like Number of Accidents or Number of Near Misses or % Staff Attending Safety Induction Training as evidence of success. But the success is evidenced by the absence of data, not the substitution of quasi-correlated data-abounding proxies!
The solution is to convert this absence of data into something more informative. You have two very simple and sensible options.
Option 1: Measure the number of days between accidents
What you compute for your measure’s value here is the number of days between accidents. The most recent measure value will be the number of days since the last accident.
A good result is when the days between accidents is larger. But with this kind of measure, it’s a little trickier to pick up signals of change easily. So Option 2 works a bit better.
Option 2: Measure the accident rate
What we do is turn our count of safety accidents into a rate of accidents per time period. It doesn’t have to be annual: the idea is to think about the rate value as opposed to the count value.
The data that you are actually collecting is the date on which the accident happened. You’ll use this raw data to calculate the rate by first working out the days between accidents, as in Option 1. Then, using the days between accidents, compute a value of the number of accidents per day by simply inverting that value (dividing it into 1). That gives you the accident rate.
Then you can use your accident rate on a Smart Chart, the absolute best kind of chart for highlighting REAL signals in your performance measures.
Find out what a Smart Chart is here: //www.staceybarr.com/webinars/YourKPIsAreTellingYouLies
Are you getting enough sense out of your measures of rare events? Find 30 minutes and a cup of coffee, and give the incident rate method a try. The Using Smart Charts course includes a template for how to convert your rare event data into an incident rate, and reveal signals by using a powerful Smart Chart.
Connect with Stacey
Haven’t found what you’re looking for? Want more information? Fill out the form below and I’ll get in touch with you as soon as possible.