Day 17: Type 2 errors
Last post we did a quick bit on type 1 errors. As with anything, there is more than one way to make an error. Today we are talking type 2 errors! They are related in the sense and we’ll go over what that means and compare the two right… now!*
Type 1 errors, as we covered yesterday are “dog alarm” or “false positive” errors. These errors are caused when we reduce the minimum confidence we need, or sometimes we have outliers. Both of these things can (and do) happen when we talk statistics. However, too much of anything can be bad and that leads us into type 2 errors.
Type 2 errors are sometimes called “cat alarm” errors. From our last post, let’s say you get tired of using the doggo as your alarm system. Instead, you are going to train the cat to be the alarm. However, one day someone breaks into your home and the cat does what cats do best, ignore the human! This is the essence of a type 2 error also called a false negative. This happens when we are too stringent with our confidence requirements.
A good example of a type 2 error would be our coin experiment, when we changed the significance level to the 5 sigma used by physicists we saw that we could not reject the null hypothesis (meaning we couldn’t say that the coin was bias). If I were someone who made bets, I would bet that in physics studies, type 2 errors happen. Maybe not frequently, but I bet they do occur and that is okay in this case because we are looking for a very, very small difference so we want to be sure that the observation is significant and not just background noise we mistake for significance.
So we’ve introduced the idea of a type 1 error, or the dog alarm and the type 2 error, or the cat alarm. Next I think we can discuss how we minimize those errors. For now, remembering type 1 and type 2 errors, might be confusing, but thankfully the internet has come through with a bunch of ways to remember the difference. My favorite is this one!
Yeah, another short (but still very important) post. We’re going to do some more diving into stats though, so the posts are going to be a bit longer. Personally, I prefer to keep them short because I think they are easier to “digest” in smaller bits and easier to write. What do you think?
Until next time, don’t stop learning!
*As usual, I make no claim to the accuracy of this information, some of it might be wrong. I’m learning, which is why I’m writing these posts and if you’re reading this then I am assuming you are trying to learn too. My plea to yu is this, if you see something that is not correct, or if you want to expand on something, do it. Let’s learn together!!
But enough about us, what about you?