4 ms·

The author is a PhD in CS from Berkeley and teaches data science at a (decent I think?) college. And yet, he writes completely incorrect statements about statis

by _cs2017_ 3y ago

The author is a PhD in CS from Berkeley and teaches data science at a (decent I think?) college. And yet, he writes completely incorrect statements about statistics. Why? > "(1) that the supposed effect is really the cause, or in this case that incipient cancer, or a pre-cancerous condition with chronic inflammation, is a factor in inducing the smoking of cigarettes, or (2) that cigarette smoking and lung cancer, though not mutually causative, are both influenced by a common cause, in this case the individual genotype." > If you think that’s the stupidest thing you’ve ever heard, you can take it up with Sir Ronald Fisher, who actually made this argument with apparent sincerity in a 1957 letter to the British Medical Journal. I mention this in case you didn’t already know what an ass he was. Of course, it's very possible that there's a common cause. For example, people who are very depressed may choose to smoke, and depression may also cause cancer. I'm not saying it does, I'm just saying it's clearly a possibility that needs experiments to disprove. The author's claim is so wrong, and yet he seems to genuinely believe it.

YokoZar 3y ago
Wouldn't we need, like, evidence or something to believe that possibility? We have mountains of direct evidence with underlining theory for the straightforward explanation. Sometimes we need to believe that evidence.
- _cs2017_ 3y ago
  No, you don't need evidence to prove that this is a possibility. You only need evidence to claim that this is impossible.
jll29 3y ago
It also seems unprofessional to call Sir Fisher an "ass", which is called an ad-hominem attack (attack on the person is obviously not a valid logical argument), and one of a dead person at that. Fact is: Fisher made fundamental contributions and the poster himself uses these on a daily basis: "Student" [Gossett]'s t-distribution, Fisher's linear discriminant, a precursor of linear discriminant analysis, ANOVA, Fisher information, Fisher's scoring, Minimum Fisher information, the F-distribution, Fisher–Tippett–Gnedenko theorem, the Fisher-Yates shuffle algorithm, the von Mises–Fisher distribution, inverse probability, Fisher's permutation test Fisher's inequality, sufficiency of a statistic, and more. His contribution to biology are even more impactful. It also happens that Fisher was a proponent of the "Frequentist" interpretation of probabilities, whereas the poster belongs to the recently more popular "Bayesian" camp of ideology (yes, ideology, if you think statistics as a part of mathematics is free from that, welcome to real life!). He further held strong views about races, perhaps because he was a professor of eugenics (which we today know as incorrect unscientific nonsense, but obviously that was not the state at the time), and there is am ongoing controversy regarding whether he was a racist or not. What I read about him suggests to me I may not have liked him, but to just stamp him as "ass" wihtout elaborating is too cheap for a statistcs professional in my view. But then this is a personal blog, where the poster can write whatever he wants under US law.
- spurgu 3y ago
  > eugenics (which we today know as incorrect unscientific nonsense How is it nonsense? I am not that knowledgeable on the topic but to me the biggest issue (outside of moral ones) is the decrease in genetic diversity. Is there anything else in particular you have "scientific" qualms with? So to me it's not "nonsense" but rather "problematic, and moreso the further you take it".

YokoZar 3y ago

Wouldn't we need, like, evidence or something to believe that possibility? We have mountains of direct evidence with underlining theory for the straightforward explanation. Sometimes we need to believe that evidence.

_cs2017_ 3y ago
No, you don't need evidence to prove that this is a possibility. You only need evidence to claim that this is impossible.

jll29 3y ago

It also seems unprofessional to call Sir Fisher an "ass", which is called an ad-hominem attack (attack on the person is obviously not a valid logical argument), and one of a dead person at that. Fact is: Fisher made fundamental contributions and the poster himself uses these on a daily basis: "Student" [Gossett]'s t-distribution, Fisher's linear discriminant, a precursor of linear discriminant analysis, ANOVA, Fisher information, Fisher's scoring, Minimum Fisher information, the F-distribution, Fisher–Tippett–Gnedenko theorem, the Fisher-Yates shuffle algorithm, the von Mises–Fisher distribution, inverse probability, Fisher's permutation test Fisher's inequality, sufficiency of a statistic, and more. His contribution to biology are even more impactful. It also happens that Fisher was a proponent of the "Frequentist" interpretation of probabilities, whereas the poster belongs to the recently more popular "Bayesian" camp of ideology (yes, ideology, if you think statistics as a part of mathematics is free from that, welcome to real life!). He further held strong views about races, perhaps because he was a professor of eugenics (which we today know as incorrect unscientific nonsense, but obviously that was not the state at the time), and there is am ongoing controversy regarding whether he was a racist or not. What I read about him suggests to me I may not have liked him, but to just stamp him as "ass" wihtout elaborating is too cheap for a statistcs professional in my view. But then this is a personal blog, where the poster can write whatever he wants under US law.

spurgu 3y ago
> eugenics (which we today know as incorrect unscientific nonsense How is it nonsense? I am not that knowledgeable on the topic but to me the biggest issue (outside of moral ones) is the decrease in genetic diversity. Is there anything else in particular you have "scientific" qualms with? So to me it's not "nonsense" but rather "problematic, and moreso the further you take it".