This is 165th #WILTW
The term ‘statistics’ strikes fear into many.
You may have 5 ‘A’ levels, have a distinction in finals, and be able to perform life saving cardiac surgery in a premature baby. It’s also likely you’ll still start sweating if a junior colleague asks you:
…why was the power calculation insufficient to reject the null hypothesis in this paper?
An extremely well shared meme was released a couple of years ago aiming to explain the often mis-understood topic of type I and type II error.
The illustration doesn’t really do the underlying principle justice but the explanation is fair.
- A type 1 error is when you decide a test result is positive, or an intervention has worked, when in fact it isn’t/hasn’t.
- A type 2 error is when you decide a test result is negative, or an intervention has failed, when in fact it isn’t/hasn’t.
I was reminded this week by Professor Mayur Lakhani of a third type of error (in fact there is also a type 4 error but I’m not going to go there). A type 3 error is when you are correct that a test result or intervention is positive but you have decided this for completely the wrong reason. Outside the realm of mathematics it is when you solve the wrong problem and don’t realise it. In an article written by Large Kiely, shared by Prof. Lakhani, this issue was explored though the lens of organisational culture:
“The type three error is a dangerous form of group-think and can happen very innocently and with all good intentions. The causes are a bit surprising, having to do with, believe it or not, too much expertise in the same field.” – Laree Kiely
In healthcare this probably occurs more frequently than we would care to admit. Take a group of highly trained consultants or managers, and set them on a problem. Very early in discussion, the cause of the problem will be decided, not via evidence, but through anecdote. Solutions will be presented to solve the perceived issues and any improvement attributed to the intervention(s). It will only be later, sometimes much much later, that other problems will materialise demonstrating the original assumptions to be flawed.
I’ve explored group-think previously but the type 3 error is a very tangible output of it.
The suggested advice by Laree Kiely is to always have 3 possible actions before deciding on one. This isn’t always easy to do but forces groups to avoid discounting relevant issues:
“Decision making studies have shown that if you think there is a right answer, then the first one that looks right becomes the final choice, and the thinking stops there. Problems today rarely have only one right answer.”– Laree Kiely
Inevitably the time required to think of a third idea to avoid the third error will mean it rarely occurs. But at the very least we shouldn’t be scared of calling it out when we make this mistake.
What have you learnt this week? #WILTW
Videos below explain the concept of type 1 and type 2 errors further