Cox’s Theorem implies that we either use Bayes or our methods will violate some simple but desirable properties. This has two consequences: (1) Frequentist methods such as p-values, which aren’t equivalent to posteriors, are guaranteed to b...
Cox’s Theorem implies that we either use Bayes or our methods will violate some simple but desirable properties. This has two consequences: (1) Frequentist methods such as p-values, which aren’t equivalent to posteriors, are guaranteed to b...
Last week I posted an old example showing how everything in the Frequentist arsenal points to while an elementary deduction from the data implies . The predictable response (an example can be found here) was to claim that no Frequentist would be that d...
Like kudzu and people who wear socks with sandals, Frequentists are going to be an unsightly part of the landscape for some time to come. Therefore I propose five principles for dealing with them. Absolute Freedom of Research: Seemingly everyone in Sta...
Objectivity and Subjectivity in statistics are often hard to spot, especially for students. Luckily however I’ve got some examples that’ll illuminate the concepts. Example 1: If a Frequentist has a fixed parameter in their model they may conduct a ...
Frequentists often claim they prefer their methods because they know how to test the model assumptions. Unfortunately, those tests don’t verify the statistical model. When a Frequentist says “the data generation process is modeled with a &#...
How could anyone have ever taken the nightmare of classical methods described in the last post seriously? It happened because on the simple problems Fisher, Neyman, Pearson et al based there work there is a rough symmetry between two very different vie...
I’m tired of the evil undead vampire that is classical statistics. I hope to drive a stake through it’s heart with an example so simple even Philosophers can understand it. I’ve been doing some reliability testing recently using expon...
The yield from the first atomic bomb was classified, and might have remained so if some knucklehead hadn’t published a series of photos complete with scales and time stamps in Life magazine. This was enough to allow the physicist G. I. Taylor to ...
Suppose there’s some aspect of our universe we’d like to know; perhaps it’s a physical measurement taken next week, or an unknown population average that exists today. Whatever it is, we use information to create a distribution which ...
Frequentism and Subjective Bayes are both special cases of Objective Bayes (to the extent they’re true at all). I’ve detailed (here, here, here, and here) in exactly what sense Frequentism is a special case of Bayes. Here I will do the same...
