♕Deborah Mayo♕

Justice demands that every individual or institution, that funded, conducted, or enabled high-risk virus research in Wuhan, along with the scientists who published those misleading papers while concealing their conflicts of interest, plus the journals that refuse to retract them, be held fully accountable. Anthony Fauci and his advisor may have been at the center of that ecosystem, but they were not alone. Many others have their fingerprints on it: researchers, intermediaries, journals, and gatekeepers who enabled the work and enforced their preferred narrative. Accountability should extend across the board. It shouldn’t stop at the surface. That means confronting and dismantling the network behind it. The public has every right to expect that.

I can only compare my getting today's word on the second round to the last time a most unlikely guess turned out to be right. 🟨⬜⬜⬜🟨 🟩🟩🟩🟩🟩

Was quite sure they'd never use such a simple word...but they did. Wordle 1,771 2/6 ⬜⬜⬜⬜🟨 🟩🟩🟩🟩🟩

Some Bayesians refuse to grant frequentists the use of any term. Frequentist error probabilities are hypothetical claims associated with methods, (e.g., the prob test T would yield d(X) > d(x) computed under h'). If they were called X-probs, some Bayesians would still say, they're not X-probs, only posteriors are. Bayesians should allow frequentists to define their terms, reject their use if you don't like them, but don't insist they must be posteriors, when they're not.

@learnfromerror @AprileBernardi @PavlosMsaouel @elmir1omerovic @StatModeling Definitely. You can't compute an error probability without a prior because you can't compute a posterior probability without a prior. Mislabeling type I assertion probability alpha as an error prob just adds confusion.

@f2harrell How would you respond to the criticism by Evans et al. in 10.1001/jama.2026.4175? Are the arguments made there against Bayesian methods in phase 3 RCTs valid in any meaningful methodological sense? In particular, what do you make of this claim: “Use of Bayesian methods in late-phase or confirmatory clinical trials has generally been limited to supplementary analyses, given recognition that their implementation can compromise evidentiary and integrity standards and the reliability of results through (1) concession of the benefits of randomization through their inclusion of external (prior) information; (2) loss of objectivity by incorporating sponsor- or investigator-specific priors; and (3) reduced robustness via reliance on strong and sometimes unverifiable assumptions.”

@f2harrell @AprileBernardi @PavlosMsaouel @elmir1omerovic @StatModeling That's exactly why a Bayesian posterior is not an error probability. But the last sentence makes no sense--a prior is the cost of an error prob???

@learnfromerror @AprileBernardi @PavlosMsaouel @elmir1omerovic @StatModeling As IJ Good elegantly stated, the exact meaning of probability is in the eye of the beholder, even for frequentist prob. P(true error | data, model, prior) = 1-P(effect | ...) so we just need to be exceptionally clear on what is conditioned on. Prior=cost of getting error probs.

@f2harrell @AprileBernardi @PavlosMsaouel @elmir1omerovic @StatModeling Degree of belief in H given e is what Bayes gives--where's the error prob? There is none. An error prob is attached to a method for reaching error prone claims based on data. It tells us how often it does so erroneously.

@learnfromerror @AprileBernardi @PavlosMsaouel @elmir1omerovic @StatModeling False. They are the only error probabilities, since they encapsulate P(making an error). Your wording about frequentist ideas is misleading. You cannot call something an error by assuming the very fact that would make it one.

@_bcalexan @f2harrell @AprileBernardi @PavlosMsaouel @elmir1omerovic @StatModeling Bayesian probs aren't error probs, can't tell just what the second means.

@learnfromerror @f2harrell @AprileBernardi @PavlosMsaouel @elmir1omerovic @StatModeling Aren’t they both error probs in a way? Bayesian provides errors probs for the data at hand and freq provides it over ‘all such decisions.’ I have read and listened to your webpage on alpha, @f2harrell. I have questions about it, maybe for another time if you’re willing.

@f2harrell @AprileBernardi @PavlosMsaouel @elmir1omerovic @StatModeling Posterior probs aren't error probs, and what's more, they're not given error probabilities. As David Cox said: "they are what they are".

@AprileBernardi @learnfromerror @PavlosMsaouel @elmir1omerovic @StatModeling You're reading the opposite of what was written. Bayes produces error probabilities directly. Frequentism doesn't produce error probabilities at all, only assertion probabilities. "My data wouldn't look like this" is a function of subjective intentions and is very indirect.

Reading @learnfromerror! 📖

Sure I can: Assume H. Derive a contradiction. Therefore H is in error.

Another example of the failure to understand reductio and falsification arguments errorstatistics.com/2026/03/26/com…

I wouldn't call you this if you excessively flatter me for getting this silly word in just 2 rounds, since it was kind of surprising, little froggy. Wordle 1,764 2/6 ⬜⬜🟩🟨⬜ 🟩🟩🟩🟩🟩

@PrometheusCHT Hope you like it! It's 30 years. What do you think of asking people jeopardy-style questions about it, giving points of value

@lakens @nulliusverbapod Which Chang? Is the reference to my name? Sorry, I don't think I know how to follow conversations on X in order.

@learnfromerror You can just let me know :) We record at @nulliusverbapod. BTW, Chang writes in the intro he wished he had been able to incorporate the ideas of more people, then mentioned your name, and all I could think was: Yes, that would have been good.

Re-reading Hasok Chang's call for pluralism in Science in his 2012 'Is water H20?'. It is so badly argued for, his discussion is so one-sided, and his rejection to engage with criticisms is so annoying, I find it hard to complete. Will discuss it in an upcoming podcast episode.

@f2harrell @PavlosMsaouel @elmir1omerovic @StatModeling Quantities that "define their own error prob" are ones that lack error probs.

@learnfromerror @PavlosMsaouel @elmir1omerovic @StatModeling Besides misusing the word 'error' you fail to understand that posterior probability defines its own error prob. (P(treatment is ineffective) = 1 - P(effective)). Your error probabilities do not make inferences and do not 'find things out', only give evidence against H0.

@f2harrell @dylanarmbruste3 @PavlosMsaouel @elmir1omerovic @StatModeling What? Where to even start!

@dylanarmbruste3 @learnfromerror @PavlosMsaouel @elmir1omerovic @StatModeling Objectivity means, for example, that we minimize the use of information that is in people's minds, and when we have to be subjective be very explicit about it (e.g., choice of prior). Completely non-objective: having to account for future data looks never carried out.

@f2harrell @PavlosMsaouel @elmir1omerovic @StatModeling You just don't how to use error probabilities to make inferences and find things out. You think inference is in the form on posterior belief assignments--but a severe tester would need to place error probabilities on those

@learnfromerror @PavlosMsaouel @elmir1omerovic @StatModeling As I've pointed out many times what you call 'error probabilities' are nothing of the sort. hbiostat.org/bbr/alpha

I don't see where intentions come in, but whenever we talk of the capabilities of an instrument, we're talking about how it would behave in applications other than this one. Yet we can use that information about its capabilities to make inferences about the case at hand. Only Bayesians think warranting an inference must mean assigning a Bayesian belief in its truth..

@dylanarmbruste3 @learnfromerror @PavlosMsaouel @elmir1omerovic @StatModeling By calling them error probabilities you are endowing them with qualities they simply don't have. What one might observe, under hypothetical investigator intentions, is a far cry from the chance of making an error once you observe something. alpha=type I assertion probability.

@f2harrell @dylanarmbruste3 @PavlosMsaouel @elmir1omerovic @StatModeling that's absurd, though I don't know what you're disagreeing about

@dylanarmbruste3 @learnfromerror @PavlosMsaouel @elmir1omerovic @StatModeling I am more than glad to disagree with Neyman et al. They did net harm to science.