Objections to Frequentism

8/14/18

(Updated 7/23/22)

Hello. 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! Remember,

  1. Prior, prior, chance on fire!
  2. Those who do not learn frequentism are doomed to repeat it
  3. I'm a Banoian

Comment from reader:

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

If you found any of this useful, please consider donating via PayPal to help keep this site going.

Disclaimer

Views expressed here are only the personal opinions of the author, and not those of any organization he is associated with.

Please anonymously VOTE on the content you have just read:

Like:
Dislike: