As a student I thought that there was no fanaticism involved in the world of Mathematics. Sure in Science you always have crackpots and competing crazy theories around but I thought such things could not possibly happen with something so aseptic and precise as math. So you can imagine my surprise when I found out about this curious religious group in the field of Statistics who call themselves Bayesians.
Bayesianism is a religion which demands its followers to use Bayes’ Theorem for any reasoning involving uncertainty regardless whether the reasoning is deductive or inductive in nature, though they also advice to consider more everyday life questions like “How would have Laplace cooked this omelet?” to then cook the omelet based on their own prior assumptions.
Bayesians regard Ronald Fisher as the Antichrist since the years of tribulations that his work brought among Bayesians souls. They see this dark period of horror as a sign preceding the rapture with the second coming of Bayes. Unfortunately they cannot agree on the estimation of when the rapture will take place since they fiercely debate which prior should be used for this important calculation but, nonetheless, every good Bayesian agrees the rapture is nigh.
Statisticians not professing Bayesianism are treated as heretic atheists and Bayesians haughtily nickname them as “Frequentists”. It is not uncommon to see flocks of Bayesians in conferences preaching the Word to unsuspected Statisticians, though their proselytical efforts to lure adepts into their coterie seem to be more effective among non trained professionals in Mathematics like Psychologists or Biologists.
Psychologists are particularly propitiatory preys since many of them study Psychology to heal themselves from mental problems; a required precondition for joining a cult. So when Bayesians approach them with open arms and cool merchandising many Psychologists can’t help but to embrace what they consider hipster statistics.
The Bayesian Holy Ten Commandments
- You shall have no other Theorem before Bayes’
- You shall not make for yourself a wrong estimator — any maximum-likelihood of anything that is in heaven above, or that is in the earth beneath, or that is in the water under the earth.
- You shall not take the name of Bayes in vain.
- Remember the Theorem, to keep it holy.
- Honor Laplace (peace be upon him) and his momma.
- You shall not use p-values.
- You shall not use confidence intervals.
- You shall not use significance tests.
- You shall not bear false priors against Bayesians.
- You shall not covet a Bayesian’s house; you shall not covet a Bayesian’s wife, nor his male student, nor his female student, nor his tenure, nor his grant, nor anything that is Bayesian’s.
If Bayesians are caught by coreligionists using p-values they are regarded immediately as gays since Bayesianism considers this approach backwards… and we all know what happens to gays in a religious environment; they become priests.
Bayesian priests themselves have doubts and temptations relieved by raping their students’ minds with warnings about those maximizing their likelihoods too much going blind. However, when temptation is too strong, you can usually find them trying to pray the p-value away in therapy groups.
Bayesians regard researchers p-hacking their results for personal benefit as victims of the evil influence of the p-value heresy, and they might often claim Frequentists’ evil ways stall research and discoveries like, for example, the cure of cancer.
Another common Bayesian accusation against the p-value heresy is the difficulty journalists have to report its meaning and, to be fair, the CNN anchor asking if meteorites are due to global warming probably agrees.
Actually Bayesians hold in high esteem journalists since they share so many philosophical points of view with them, for instance, just as journalists sing the song “Objectivity does not exist” when anyone mentions the word “lie”, so do Bayesians when anyone mentions the word “prior”, and just as journalist don’t let facts spoil a good story, Bayesians don’t let data spoil a good experiment.
Today is the anniversary of Sir Ronald Aylmer Fisher’s birth, and I thought I would commemorate it by humorously describing the “fallacious rubbish” religion he hammered nearly a century ago. Happy birthday Ronald! ;)
- Economists argue about Bayes (andrewgelman.com)
- Reverend Bayes takes Sir Ronald to the mat! Wait hold on…that might have been Father Mendel!! (ecologicallyoriented.wordpress.com)
- Closet Bayesian (johndcook.com)
- When Fen-Dwelling Bayesians Can’t Handle the Truth (bactra.org)
- Frequentists vs. Bayesians (and a comment about the optimal alpha approach) (ecoloblogy.wordpress.com)
- Bayesian Inference for LDA is Intractable (lingpipe-blog.com)
- Bayes’ Theorem (semedx.wordpress.com)
- Introduction to Bayesian statistics (freshbiostats.wordpress.com)
- The Monty Hall Problem As An Introduction to Bayesian Reasoning (thinkthatthrough.wordpress.com)
- Is Bayesian Inference a Religion? (normaldeviate.wordpress.com)
- Don’t let the facts spoil a good story (guardian.co.uk)