Objections to Frequentism


(Updated 8/6/19)

8/14/18


Hi. I'm a use whatever works type of mathematical statistician: frequentism, nonparametric, machine learning, Bayesian, operations research, SAS, Python, Excel, whatever. However, after reading, hearing, and collecting many arguments against frequentism, some of them contradicting, and most of them containing fallacies rehashed year after year, I decided to write this article in response.

Frequentism is roughly defined as:


Criticisms of Frequentism:


Responses to Criticisms of Frequentism:


Frequency definition of probability

Frequentist concepts like hypothesis testing and p-values are too difficult to teach.

Frequentism relies on data you didn't observe.

Frequentism is bad for science.

Fisher dislike or envy

Frequentism is used to p-hack.

Frequentism is responsible for the "replication crisis".

Just look at all those counterexamples to frequentism!

Counterexamples, paradoxes, or issues in Bayesian probability and statistics

Frequentism is really just a special case of Bayesian.

But everything is subjective anyway!

What is your background?

Thanks for reading and sending in comments/corrections!

Comment from reader:

I got many answers in the frequentist vs Bayesian thread, but this pro-frequentist page is really superb!




If you enjoyed any of my content, please consider supporting it in a variety of ways: