Friday, November 4, 2016

Understanding 'Error'

Within decision psychology there’s a famous experiment, described in Daniel Kahneman’s book, Thinking, Fast and Slow, known as the Linda the Bank Teller experiment. The experiment goes like this – participants are given the following information:

Linda is 31 years old, single, outspoken and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations.

Participants are then asked the following question (see which answer you would choose):

Which of the following is more probable?
    1. Linda is a bank teller.
    2. Linda is a bank teller and is active in the feminist movement.

If you chose #2, then you are like most people. Most of the participants in the study identified that it is more probable that Linda is a bank teller and is active in the feminist movement than her being simply a bank teller.

However, most people (and you, if you chose #2) are wrong. The ‘correct’ answer is #1. Here’s why – the experiment asked you a math question. Most people don’t necessarily see it that way, but that’s what it is. They asked about the probability of an event occurring. And the thing is that you cannot have the probability of two events being greater than the probability of one of those events occurring.

To put it another way, the question the experiment asked you is like asking you:

Which of the following is more probable?
    1. That you went to work today?
    2. That you went to work today and you had a cup of coffee?
At best, the two probabilities can be equal, but #2 can never be greater than #1.

So, if you, like most participants in the study chose #2 you made an ‘error’. This ‘error’ you made is one that is predictable and, according to Kahneman and others, is an example of the irrationality of people. The message that is often told is that people are unreliable and can’t be trusted.

‘Human error is a common issue in many workplaces, and is a big problem that many safety professionals actively try to deal with. Some people believe that ‘human error’ or ‘unsafe acts’ (or whatever we want to call it) is responsible for most of the accidents we have in the workplace. So we spend a lot of time in safety trying to understand and deal with ‘human error’, in all its forms and studies like the Linda the Bank Teller study seem to confirm our suspicions – people are a problem to control.

But wait a minute…if you chose #2, why did you choose #2? If you’re like most it’s probably because of all the additional information that was given to you. You used a stereotype, or a mental shortcut. Your mind filled in the blanks of Linda’s life based upon the information you had. Why does your mind do that?

Well, as psychologist Gerd Gigerenzer points out, it’s because this mental shortcut is correct a lot and allows you to communicate more effectively and efficiently with other people. Think about it, if you listen to the words people typically use in normal conversations there’s a lot of ambiguity. We often aren’t very specific about what we’re talking about. So our mind fills in the blanks, using the information that is said, as well as other contextual factors. And this works most of the time. It allows you to carry on conversations with others without a lot of confusion. Effectively, this makes you more socially intelligent.

Now wait a minute, none of the above about social intelligence makes choosing #2 any less of an ‘error’. But when we try to understand the ‘error’ we find that this ‘error’ makes people more successful overall in the sorts of environments they operate in. To put it another way, the ‘cause’ of the error in this experiment is the ‘cause’ of success most of the time.

And this is a common theme we see when you start to dig deeper into human performance – success and failure have the same causes. People have to deal with a world that is imperfect and complex, all the while with scarce resources. To make this work people make what Erik Hollnagel calls performance adjustments. These performance adjustments are remarkably successful most of the time…until they aren’t.

What does this mean for the safety professional? There’s a few lessons learned here:
  1. We don’t want to eliminate the causes of ‘human error’.  If people fail and succeed for the same reason, the performance adjustments they make, then eliminating their ability to adjust their performance will not only eliminate the errors, it will eliminate your successes as well. And this is not simply production success, but also safety success. Often these performance adjustments are the reason you are having a lot less accidents in your organization than you might otherwise be. As Steven Covey said, “seek first to understand”. We want to understand ‘error’ before trying to eliminate it.
  2. ‘Human error’ is never simply an issue with the ‘human’. Because success and failure have the same cause, this means that the error you see is always only part of the issue. If people are making performance adjustments and this is what is causing the ‘errors’ you see, then you have to ask what they are adjusting their performance to. Those who try to deal with ‘error’ by focusing on the individual are dealing with this problem with one hand tied behind their back while blindfolded. Focus on the context of the performance adjustment you see and you’ll begin to understand why it made sense for the person do to what they did.
  3. Focus on what makes people more successful, not merely on what makes them fail less. Because these performance adjustments are tied to people’s attempts to achieve success and avoid failure, telling them to try harder, pay more attention, or be more mindful is unlikely to work. Instead, look for ways to make successful performance adjustments easier. For example, if we framed the Linda the Bank Teller problem by telling you exactly how many bank tellers there are and how many of those bank tellers are active in the feminist movement, it is less likely that you would have made the same mistake you did. By contrast, if we tell you to simply pay more attention next time, it’s likely many will make the same mistake again in the future. By creating the conditions where people can be more successful you enable people to be more successful in a sustainable way.

If you have questions about how you can create sustainable performance adjustments in your organization contact us today! 

1 comment:

  1. This whole idea of bias is problematic. Gigerenzer’s views on this matter offer a valuable counterbalance. Your thought about focusing on what makes people successful is something that Gary Klein promotes in his book, sources of power. John Flach and Fred Voorhorst have some useful things to say about the problems with bias in their book, What Matters? Excellent book. Free download at

    One thing worth noting is that the Linda problem has to be tuned very carefully to get the effect. A word or two wrong in the experimental materials and the bias disappears. That is typical of the experimental method in this area. If you want to shine effect in this area, you have to work hard at getting your experimental materials set up properly.