What are type I and type II errors? They sound fancy but are quite simple to understand and are a good way to look at solutions to problems. The easiest way of thinking about type I and type II errors is as false positive or false negative results to a test.
You can probably think of plenty of examples for this: pregnancy tests, covid tests, etc. But let’s look at an example on the macro scale, an aircraft. After 9/11 governments realised that aircraft were susceptible to terrorists (or any intruder) being able to enter the cockpit of the aircraft. This is a bad thing. So they come up with a way to stop this, a secure door.
So our aircraft now has a secure door to keep intruders out. We are now secure, right? Not quite, there is an insider risk – a nefarious pilot could now lock the other pilot out of the cockpit and crash the aircraft, this occurred in the German Wings disaster in 2015. If the secure door is our security test – that is: if the door is secure, the aircraft is secure – this is what we call a false positive. We thought we were secure, but we were not.
This table demonstrates the type I and type II errors in our security test.
| Hypothesis: if the cockpit door is secured to prevent unauthorised entry, the aircraft is secure. | Possible outcomes | ||
| Aircraft is secure | Aircraft is not secure | ||
| Test result | Door is secured | True (intruder locked out) | Type I, false positive (pilot is nefarious) |
| Door is not secured | Type II, false negative (nefarious pilot had no opportunity or no intruder at all) | True (intruder enters cockpit) | |
There you have it, you now know the basics of type I and type II errors and have a real world example to explain to your friends. So what is the point of this? The next time you are trying to find a solution to a problem, make yourself a table like the one above and insert your hypothesis, test conditions and possible outcomes. Then see if the result makes sense, it will help you to identify if you are actually solving the problem or just have a good idea.
Reflection
This was quite the journey getting my head around this. I questioned whether I should change my example, maybe I was holding onto it because I’d already put effort into it, or my interest in aviation was making me hold onto a bad example. But comments from classmates indicated the context was valid and I had that feeling I get, like a mental version of the Necker Cube, that I just wasn’t quite looking at it right. So I pushed on.
This is a useful tool for looking at solutions and problems, to help identify acceptable risk and also if you are looking at a solution that actually addresses the problem, or even looking at the problem.
References/background reading
https://en.wikipedia.org/wiki/Type_I_and_type_II_errors
https://www.scribbr.com/statistics/type-i-and-type-ii-errors/
