The need to rethink statistical education.
The ways statistics are taught in many universities is detrimental to science
I originally planned to write an article about Bayesian vs. Frequentists Methods. Midway through my writing, however, I realized that the issue isn’t really understanding the difference and then “choosing” aside. Both schools of thought have good reasons to be applied to one setting or the other. In some cases, these seemingly different ways of viewing data and the world are not even that different.
Reading through some articles and books about Bayesian Methods made me realize how useful these could be, but also how little I learned about them in my university stats education. I learned about Bayes Theorem and how you can use it in clinical settings (the famous disease/disorder and test examples). That’s it. And I thought that was it for quite a while. I was not aware of the broad class of methods that could be derived from this theorem.
I was also not aware that most of the “classical” methods I learned, like t-test, ANOVA, Linear Regression etc. are closely connected and are all instances of the General linear model. Our professor told us that in the last lesson we had.
He even tried to connect the dots and show the unity among the vast array of methods we had met as separate entities. But…
