Take the next step to accident prevention, safety expert urges
December 1, 2013
KITCHENER, Ont. – The rising cost of lawsuits, increasing government scrutiny and penalties for unsafe operators, and easily accessible technology that can turn any bystander at the scene of an accident into a star witness can mean only...
KITCHENER, Ont. – The rising cost of lawsuits, increasing government scrutiny and penalties for unsafe operators, and easily accessible technology that can turn any bystander at the scene of an accident into a star witness can mean only one thing for an industry as visible as transportation: Bad decisions when it comes to safety cost a huge amount of money these days.
That’s part of the reason why motor carriers need to move beyond being incident-reactive to being prevention-proactive, according to Mark Skinner of the Infrastructure Health and Safety Association. For Skinner, who spoke at the Fleet Safety Council’s 22nd Annual Educational Conference, that means moving beyond mere compliance with safety regulations towards predictive behaviour analysis – the science of analyzing past behaviours to predict the likelihood of such behaviours occurring in the future. In the context of safety, it’s figuring out which behaviours can lead to accidents and taking action before the accident occurs.
“If all we are doing is compliance, we are not really helping anyone,” Skinner said. “Predictive behaviour analysis is forward thinking, and I mean really forward thinking. But it’s not magic; it doesn’t just happen. It’s a lot of hard work.”
Fortunately for motor carriers keen to take the road towards predictive behaviour analysis, all the data made available by technologies such as the engine control modules recording hard braking or rapid acceleration, provide a rich source of data to be mined.
So can simple observations such as how drivers get out of their cabs – do they use the proper three-point technique or leap out on a wing and prayer?
“A lot of companies may want to do predictive analytics but have yet to master basic recording,” Skinner warned.
Observations of safety-related behaviour should be both frequent and plentiful – at least five or six times per week – he advised. Record the number of safe behaviours observed, the number of at-risk behaviours observed, and apply a severity grade by calculating the percentage of risky behaviours observed that are of medium severity or above.
Take nothing for granted in your observations, Skinner said. For example, don’t lull yourself into a false sense of security just because you are providing safety training to employees. Just as important as how many people attended the safety meeting and what you taught them is what information they actually retained. Do you test them afterwards to determine if the safety message sank in? Do you follow up to ensure those safety-related behaviours are still being practiced months later?
Skinner also provided a checklist to follow to create a basic score card.
1. Identify the objective (for example, no lost time injuries) and gather the relevant historical data;
2. Build the scorecard by identifying the factors that will influence the objective; determining the weight of these factors; and assigning point values. He advised that point values are assigned intuitively – it’s not an exact science;
3. Develop a segmentation scheme so that the project can be broken down into digestible parts rather than one huge endeavor that overwhelms the organization. Skinner advised identifying three to six different segments;
4. Develop the baseline by documenting current and historical performance of each segment identified;
5. Compare the results of the model with current performance. Sometimes it’s important to test different versions of the model to achieve desired results.
Such models can be developed in-house on something as simple as an Excel spreadsheet program. But if you want more sophisticated and dynamic modeling, going to a third-party provider such as Deloitte Analytics would be best.
“In cases where predictive analytics is applied all the time, it works,” Skinner emphasized. He provided the example of Cummins International’s focus on its distribution branch. It collected a great deal of data: more than 210,000 audits, more than four million observations of which slightly more than 229,000 were deemed to be observations of “at risk” actions or behaviours. The company was able to correct more than 224,000 of those.
“Give predictive behaviour analysis serious consideration. It’s the way of the future,” Skinner concluded.