Type 1 vs. Type 2 errors and Stats Nonsense

Type 1 vs. Type 2 errors and Stats Nonsense

I recently had a discussion with some students on avoiding Type I and Type II errors in their research. However, it soon became clear that after some discussion, no one in the room knew what theses errors were, and upon further digging, most of the terms many scientists talk about e.g. p-values, central limit theorem, ANOVA, etc., are just that: talked about, not understood.

This isn’t a new problem, in that we all “fake it ’til you make it,” but I think we should start pushing a much more sensible narrative for new grad and undergrad students: it’s okay not to know everything, and if you don’t know, find out. I would say ask, but if you’re a fourth or fifth year grad student asking your advisor what a p-value is, there may be some social consequences, yet this shouldn’t scare you away from finding out about what these things mean. Swallow your pride for a second and learn about it, because it’s much worse to find yourself in a situation, say an interview, where they ask you a basic question from your field, and you struggle to piece it together.

Blaming doesn’t solve much, but what I will say is that another contender for adding to the trend of claiming you know certain terms, but don’t quite understand them at their core, is the advent of software that does everything for you. Now, I’m not saying we should go back to calculating ANOVAs by hand or by using old text books to look up z tables, but when training your students, teach the theory and root of the concept, don’t just have them punch it into SPSS or use an old R script to get the job done.

For peace of mind: type 1 errors are false positives, meaning that you THINK some effect is there, but there isn’t, and a type 2 errors are false negatives, meaning that you THINK some effect is NOT there, but there actually is.

Anyway, I guess my point is that yes, sometimes you gotta fake it until you make it, but it’s okay to spend some time each day checking in with yourself to see if you’re really confident about the terms you sling around all day. And, if you find yourself caught red-handed in an interview or meeting, simply admitting that you don’t know and are willing to learn is gonna be better for you and everyone around you.

Stay scientific!


Leave a Reply

Your email address will not be published. Required fields are marked *