iiinova — where your digital self keeps evolving

Boris Sobolev @sobor.bsky.social · 26/01/2025

For the same reason trialists are so excited about not knowing treatment mechanisms: they want to get away with ATE whereas patients expect clinical decision-making based on ITE.

110

Вечер перестал быть томным.

000

@pwgtennant.bsky.social Peter, your repost should bring us who study SES in health to re-evaluation of how we do research. We should focus on causal questions, causal methodology, DAGs, probability of benefit, recovery from selection bias, transportability of effects, causal fairness.

021

hold on, hold on, we’ll get to foolishness in a moment. In Fisher 1926, you read ‘averages’. A question for an undergrad completed an elementary course in statistics — To which group of statistical objects do averages belong: estimators or estimates?

100

Great! I’m glad you agreed it’s ‘concluded’… That’s right, it is not just an estimator.

100

Indeed, this should be engraved at the entry to every EBM office: 'Randomization ensures a valid error estimate. This may be applied to test the significance of observed difference btw averages of the treatment groups.' Not a word about causality, as it is concluded by reasoning, not statistics.

100

Oh, by all means, please test the significance of the observed difference between the averages of plots treated differently. 😂 Average here, average there.

100

Probability of benefit, PNS, is the future that personalized medicine is going to discover…

100

According to Fisher, it’s about ‘testing observed differences between average outcomes of groups treated differently’

100

Fisher said many things, including about eugenics and tobacco smoking. But he didn’t say that thing you attributed to him.

200

I am not. I am an arrow. I am a straight line going beyond the EBM horizon.

100

On the right, I see an estimator, which you reason is unbiased due to design. Fine. What is the estimand? What is it estimator of?

100

again, what makes it causal? all the probabilities here are conditional 🤷‍♂️

100

If one can’t explicitly demonstrate the target quantity of measurement, how do they know they measure it? That tells all about the scientific pretensions of EBM preaching

100

Like we say in Russia: на воре шапка горит… 🫣 agreed, the preaching of EBM High Priests is totally useless in determining causal estimands

100

A sleight of hand, again? ay-ya-yay… Do they, in experimental design, teach you machinations? Just put it down, in math notation or layman words, the causal estimand of an RCT… not reasoning, not statistical phraseology, not EBM blurred vision, not Rubin’s charade; just an estimand… then we talk

100

Are those the same books that make you struggle with writing down a causal estimand? It reminds me an old joke about Communism: ‘We promised you the bright future; no one was promising food and shelter’

100

I can recommend a great optometrist.

100

It’s a free country, you can fancy any name. But that’s just a name in your head. Cause-and-effect relationships is a feature of this world. Causal inference is how we convince ourselves and others that changing X will change Y.

110

I thought we already agreed that ‘sleight of hand’ is not admissible in our debate 🫢 My example of a random function gives you several values for the same element from the probability space and shows that your animus toward ‘counterfactuals in causality’ is simply misguided…

100

‘very’ bad? is the adverb really necessary? are you saying you can’t define a family of random variables on the same probability space indexed by t from the value set T? Perhaps some elementary textbook of probability can be useful… hé hé

100

Well, math makes no sense to those who don’t know it, right? For the rest of us, a random function would suffice… no counterfactuals, no strawman, no crows, no windmills… 👇🏻

100

oh, wait a sec… are you arguing with Pearl here? or, just pulling of the ‘strawman’ argument? it is kinda lame… I thought we were having a genuine debate: you and I, man to man…

100

ah, you transitioned from math notation to pics 😊… ok, let’s do a pictorial argument

200

Who said anything about counterfactual here? 🙄 “En resolución, él se enfrascó tanto en su lectura, que se le pasaban las noches leyendo de claro en claro, y los días de turbio en turbio; y así, del poco dormir y del mucho leer, se le secó el cerebro de manera que vino a perder el juicio.”

100

I’m glad you asked!First, ‘observational’ refers to the method of assessment, right? Since John Stuart Mill, we define individual causal effect in unit u as Y(1,u) - Y(2,u), where Y(x,u) is the state of unit u treated with x. Huzzah!

120

Sleight of hand, my friend? 😂 Statistics and communication aren't the same either, yet you're happy to use H0 symbol and the concept of conditional expectation. Humans are good at causal epistemology, because we observe and experiment, conceptualize and develop symbols.

200

Symbols and notations is what humans use when they pass knowledge… 😂

100

Boris Sobolev @sobor.bsky.social · 20/01/2025

Hypothesis testing about statistical estimates is the domain of statistics. Causal inference is concerned with deriving estimands of causal effects. Can YOU give a rigorous notation of causal estimands using statistical symbols and concepts?

100

sorry, what’s ‘this’?

100

Boris Sobolev @sobor.bsky.social · 19/01/2025

to keep you away from turning 65

Boris Sobolev @sobor.bsky.social · 18/01/2025

Михайло Ломоносов говаривал: "На немецком надо говорить с врагом, на французском - с любимыми, на итальянском - с Богом, на английском - с другом. А на русском можно говорить со всеми." А про сложности с наблюдениям - на испанском?

100

Boris Sobolev @sobor.bsky.social · 18/01/2025

цели ясны, задачи понятны - за дело, товарищи!

100

Boris Sobolev @sobor.bsky.social · 18/01/2025

ATE? nah… We ask how many patients would get better with treatment but not without it. youtu.be/ms4z2i7JzVY

youtu.be

Probability of benefit in clinical practice

YouTube video by SobolevSpaces

Boris Sobolev @sobor.bsky.social · 17/01/2025

Oh, that’s easy. Y⊥X|experimental method I’m glad you asked

100

Boris Sobolev @sobor.bsky.social · 24/12/2024

sometimes, a flower is just a flower

Boris Sobolev @sobor.bsky.social · 23/12/2024

Un laberinto de tentaciones

000

Boris Sobolev @sobor.bsky.social · 21/12/2024

If you've never tasted the dishes prepared by la chefa Anayeli Suazo of Tentaciones, you don't know what a gourmet meal is and what it looks like.

000

Boris Sobolev @sobor.bsky.social · 20/12/2024

Nothing can beat sleeping until noon on vacation

Boris Sobolev @sobor.bsky.social · 14/12/2024

I had a quick chat with Elias Bareinboim on the way from the a/p #neurISP24… He reminds me of Mozart, who barely has time to write down all the revelations of causality pouring on him from above. 😊

020

Boris Sobolev @sobor.bsky.social · 23/11/2024

Wondering about #causal mediation analysis in comparative effectiveness research? Check out our framework for the mediating effect of treatment time of 2 treatment methods for complex coronary artery disease. doi.org/10.9778/cmaj...

Boris Sobolev @sobor.bsky.social · 20/11/2024

There’s this misconception that 'causality is in the design.' Apart from being uninformed, this assertion is fed by a bigger fallacy: that causality is a product of our mind rather than a feature of the material world. Pearl’s major contribution is grounding causality in materialism.

032

Boris Sobolev @sobor.bsky.social · 20/11/2024

Oh, I agree that causal claims come from reasoning, and that design is a part of it. I just don't see how reasoning using words can be superior to reasoning using mathematical notation. Pearl’s formulas easily describe RCT. But trialsts have a hard time to describe the estimand of their experiment.