“If it wasn’t for bad luck, I’d have no luck at all.”  — Walter Mosley

Have you ever known someone who seemed to have more than their share of incidents? Ever worked with such a person? Supervised such a person?

The idea of an “accident-prone” worker poses some special challenges, especially in a workplace where the consequences of an incident can have severe impacts, such as in the chemical process industries.

What Does “Accident-Prone” Mean?

We use the term “accident-prone” pretty loosely. Certain models of cars have been called “accident-prone”, as have certain stretches of roadway. Here, though, we’re talking about people. The term gets used two ways: people who have more than the average number of accidents or people who are predisposed to have accidents because of their personality traits. The first is a statistical definition, the second, a psychological definition.

The State of Maine has a statute that allows for a reexamination of “accident-prone” drivers in order for them to keep their driver’s license. The license can be revoked if they do poorly on the reexamination. The statute defines an accident-prone driver as one “who has contributed to the cause of three or more accidents within a period of three consecutive years.” A statistical definition.

An acquaintance shared that a chemical plant where he works considers someone “accident-prone” when they have accidents at more than twice the average accident rate at the plant; another statistical definition.

The problem with statistical definitions is that they cannot be applied until after the incidents have happened. Since the goal of safety professionals is to prevent incidents, it would be really helpful know that an individual is “accident-prone” before they are involved in an incident, and to design strategies accordingly.

Enter the psychologists.

Psychological Testing for “Accident-Proneness”

There are psychological tests available today that claim to look at how “accident-prone” someone is. These tests look at personality and attitudes, as well as behavioral factors.

The concept of certain people having a predisposition to incidents was first formally introduced shortly after World War I by two British researchers, Major Greenwood and Hilda Woods. Their work was the basis for much that followed, but it wasn’t until World War II that the idea of accident-prone really took off.

Following World War II, the editorial board of the American Journal of Public Health published an opinion piece that cited several recent articles on the phenomena and shared the anecdotal evidence of a trucking company that removed its “accident-prone” drivers and therefore cut its accident rate by 80%. The editorial made a point of saying that being accident-prone was not the result of clumsiness or absent-mindedness, but was the result impulsiveness, rebelliousness, and resentment—the attributes of an angry adolescent.

For safety professionals, however, the angry adolescent explanation is not much help.

Statistical Significance

Studies looking for an explanation of what makes a person “accident-prone” must rely on identifying those individuals after the fact and then looking for causes and factors related to that characteristic.  If that is as far as it goes, though, then it is mere correlation. Just recently, a study published by Uplift Legal Funding found that people named Kyle and Hailey are more likely to be involved in incidents than people with any other name.  Obviously, some name must be at the top of any list of names that is compiled. But this doesn’t explain anything. While the press had a field day with the findings, only the most superstitious among us would refuse to hire someone named Kyle or Hailey for fear of their accident-proneness.

There are also less obvious cases of correlation being confused with causation. A study at a refinery complex showed that 3% of the workforce accounts for 22% of accidents. Is that meaningful?  Imagine a workplace with a workforce of 1000 in which one significant incident occurred during the study period.  In that case, 0.1% of the workforce accounted for 100% of the workplace incidents. This is an extreme example of that kind of statistic, yet no reasonable person should conclude that because one person was involved in an incident that person was particularly accident prone. All incident statistics are going to show that a relatively small portion of the workforce accounts for most incidents; even in the most unsafe workplaces, most employees aren’t involved in incidents at all.

If we are going to accept “accident-prone” as real phenomenon and act on it, then we must have a good definition of “accident-prone”. Rather than “people who have more than the average number of accidents”, the definition must account for the statistical significance of the difference. Before labelling someone as accident-prone, we must disprove the null hypothesis that their accident frequency is within the statistical distribution of accidents, the Poisson distribution:

PR(y) = (eήy)/y!

In my tattered copy of Statistics for Experimenters, the authors point out in one of their examples that with a mean frequency of incidents, ή, of 2.1, the probability of the number of accidents, y, of 5 or more is over 6%. They add, “We must remember of course that this will be a chance phenomenon and will occur not because she is less careful in these years.”

Circumstances

Recent research seeking to validate the notion of being “accident-prone” finds that studies attributing accident-proneness do not hold up to statistical scrutiny. The results, when they do find correlations, are inconsistent. Not much of a basis for a safety strategy.

Instead, psychologists have begun to shift away from personality as an explanation.

Rather than thinking of certain employees as having a psychological disposition toward incidents, with traits such as anger issues and overconfidence, psychologists are now leaning toward an explanation based on circumstances.

These circumstances include distraction, lack of sleep, emotional stress, and being under the influence of medications or intoxicants. So, rather than looking to a psychological profile, safety professionals should look to the circumstances at the time. So, situational awareness applies not just to the person doing the work, but to the people supervising the work.

Is ”Accident-Prone” Real?

As convenient as it would be to blame incidents on someone being “accident-prone”, it is not really much help. Assuming that it is a real thing, rather than a superstition, we do not have the tools to identify in advance who is and is not “accident-prone.” Even retrospective accident histories will fail us because we are unlikely to have enough history to draw statistically significant conclusions.

Since its formal beginning, observations of being “accident-prone” have largely been an artifact of the statistical distribution. It’s an appealing idea, but it is not much more than victim blaming.

What to Do with Accident-Prone Employees

If there is an employee that coworkers believe is “accident-prone”, the most important step is to disabuse everyone, especially the labelled employee, of that corrosive notion. Once someone comes to believe that they are accident-prone, it can become a self-fulfilling prophecy.

The best way to reduce the frequency of incidents is to understand what causes them, and then tackle those causes. Blaming an incident on someone being “accident-prone”, for whatever reason, simply becomes an