💾 Archived View for dioskouroi.xyz › thread › 29368960 captured on 2021-11-30 at 20:18:30. Gemini links have been rewritten to link to archived content
-=-=-=-=-=-=-
________________________________________________________________________________
This reminds me of the sort of light poking of common "correlation is not the same as causation" and "beware of confounding factors" statistical failures behind the Church of the Flying Spaghetti Monster "clearly reduction in pirates has caused global warming!" [0]. But it's a major in modern public discourse, and one for once that I'm quite willing to lay heavily at the feet of the public education system. Easily one of the most valuable classes I took in my entire time K-12 was AP Probability & Stats as a sophomore, but that was an entirely optional class with restricted openings anyway (a single teacher in a school of 1200+) which the vast majority never took even where it was offered. Yet interpreting the deluge of data in the modern world requires some level of being able to reason about things probabilisticly and have a sense of what actually goes into a measure of "significance", null hypothesis and the danger of result-driven analysis finding links that don't exist and/or cannot possibly be causitive, how sampling a population works and what the error bars look like at different sizes, random vs biased distributions, the underlying distributions, confounding factors etc. I really wish in general kids got started on some light probability thinking as early as possible, in elementary school even, without any real math (let alone calc and such) yet, but just some initial stuff to start to illustrate the mindset. Lots of very fun games and hands-on exercises with dice and so on use or could be made to use important aspects of probability and its misuse.
At any rate, I'm certainly not an expert. But there seems to be some missing BS filter where people can recognize something as silly if the example is silly enough but not in the exact same logic fail for something that seems "more reasonable somehow".
----
0:
https://www.spaghettimonster.org/wp-content/uploads/2007/10/...
Statistics are meaningless without a rigorously examined causal model of the phenomenon under investigation. In my experience of statistics education, the art of crafting causal theories was scarcely addressed.
This is very true. Systematic bias in observational data collection (astronomy etc.) as well as systematic bias in experimental data collection (particle physics etc.) isn't accounted for in statistical analysis of that data.
A classic example I recall is a Feynmann story, where a group of researchers were getting very statistically sound and repeatable results of very unusual and unexplained particle track behavior in a cloud chamber. Feynmann looked at the data and said "you probably have a tiny piece of metal in the cloud chamber somewhere" and that turned out to be the explanation.
Similar examples in the social sciences include systematic bias in the preparation and administration of IQ tests to different groups of people (see Charles Murray's 'Bell Curve' vs. Stephen J. Gould's 'The Mismeasure of Man').
Hundreds of other examples can be found across all scientific disciplines, unfortunately. To quote the smartest PI I ever worked for "There's a lot of BS in statistical analysis".
I don't get this attitude, if things are correlated, it should at least make a scientist wonder why. It certainly could be random chance, but correlation can also lead to establishing a causal model or discovering a third variable. If two things keep happening in conjunction, it at least merits further investigation.
It seems like there's this extreme reaction against people behaving like correlation equals causation, but instead of over-emphasizing correlation, it gets dismissed entirely.
Yes, the classic example being the connection of lung cancer to smoking. Initially it was just a correlation, but the correlation encouraged scientists to see if there was actually a causal relationship (which of course there was). Yes, many (probably most) correlations are spurious, but the existence of a correlation is very useful for scientists looking to find a hypothesis to test.
I share this criticism. It's almost like you have to scream at people that although correlation doesn't imply _direct_ causation, it most certainly does imply some causal chain.
Ruling out chance through significance and power, what phenomenon in the world is correlated without being causally related _somehow_?
Take the worst intersection in the country this week for traffic accidents. Close the roads to the intersection at 3am and organise a contemporary expressive dance performance to rid the intersection of its evil and re-open it 30 minutes later.
Repeat every week with the worst traffic intersection for accidents.
What you find is. IT WORKS! HURRAH! These intersections are more than usually not the worst the following week! The evidence is clear. The correlation is utterly compelling. It is significant. It has power. How could it possibly be unrelated?
Now if we stop it being comically silly in our example and make it a red light & speed camera, see how the issue is much more difficult. There is a clear line of potential causation of fixing dangerous intersections. But is it really better than folk dancing for 20 minutes at 2am? [1]
[1] This example should not be interpreted as being an opposition to all red light and speed cameras.
Well, it can also be pure randomness
They did say “ruling out chance”.
There is a scale of certainty with causality. Evidently we aren't always good at expressing the difference between light evidence of causality and heavy evidence of causality. Do we have the right words? Are we using them?
So in other words, several of the commenters here are right. On the one hand we shouldn't jump to conclusions, but on the other hand we should listen to the clues.
Consider that we can’t even truly conclusively identify true causation using scientific experiment alone. Correlations and conjectures is ultimately all we have.
Nevertheless, the fallacy is so commonplace. You will easily find a seemingly educated person selectively balking at the notion that causal relationship is ultimately a conjecture _or_ at the notion that causal relationship is possible, depending on their pre-existing beliefs.
Being emotionally attached to purported causal relationship X->Y, they will count all correlational evidence in favor; when pointed out that the evidence is correlational, they will wave it off with more correlational evidence.
If that causal relationship does not happen to align with their world view, of course, they will be right onto you with the old correlation-does-not-imply-causation mantra.
> if things are correlated, it should at least make a scientist wonder why
There's billions of possible variables. There's N^2 possible pairs of variables. It's not feasible to look at every pair that is correlated.
> if things are correlated, it should at least make a scientist wonder why
Not really.
See:
https://www.tylervigen.com/spurious-correlations
You should still be curious as to why those correlations exist. It's still important even when the reason is p-hacking, since then you can identify dubious statistics.
yep, once in a while, it will end being a once in a life discovery.
so, everything defying common sense should be reviewed, to find why and how exactly is this happening.
I don't remember who said that (pretty sure that it was R Monroe in xkcd, as usual): correlation is not causation but the data is heavily winking at you (it was more funny in his words)
The average citizen doesn't need to craft casual theories, they need to be able to look at them critically. Yes, an introductory statistics course doesn't cover everything, but it's absolutely the 20% effort that gets 80% of the results. Most of the worst misinformation I've seen lately surrounding covid, vaccines, etc, would be solved if everyone had a basic understanding of introductory stats as currently taught.
If we can get there, then we can talk about what we can do to improve things from there.
> The average citizen doesn't need to craft casual theories, they need to be able to look at them critically
Yes, I agree, and this is precisely my point. My experience with introductory stats was a heavy focus on the technical details, when in fact what would be more effective is focusing on _statistical logic_.
I’m specifically thinking that something like Judea Pearl’s _The Book of Why_ would be good to introduce early in stats education.
I think the biggest thing is whether people have a desire to dig deeper or not. I seems there are many people who just want to believe what they are told as long as it matches their beliefs.
Anything else is harder - and for many people, impossibly hard. Always has been this way, likely always will be.
Most people just want to pay the bills and put food on the table.
I think you underestimate the fact that most of the people we are talking about are more accustomed to "magical thinking". Any stats education that didn't align with their core beliefs would be dismissed as wrong. Even if you could proof out everything and they had the core intellectual horsepower to understand there is a level of belief in belief that underlies their worldview that you can't overcome.
I think I agree. Instead of stats I would propose critical thinking. Critical thinking was one of the more engaging classes for me. It covered a vast array of fallacies and how they can be exploited in real world. I am not sure if it is a common requirement now, but it certainly should be.
> Most of the worst misinformation I've seen lately surrounding covid, vaccines, etc, would be solved if everyone had a basic understanding of introductory stats as currently taught.
I don't know, most of the misinformation on the matter I've seen is just flat out wrong. Not misinterpreting statistics incorrect, just flat out lying, using false numbers or statements, etc. Having a cursory knowledge of statistics won't help you if you're incapable of Googling to check if a statistic is true or not.
"Most of the worst misinformation I've seen lately surrounding covid, vaccines, etc, would be solved if everyone had a basic understanding of introductory stats as currently taught."
I think a lot of it is more ideological at the "average" citizen level - on both sides. Most people aren't looking at the data, they're believing whatever they're told that aligns with their personal beliefs.
Can you point to a good source on that? The stats books that I've seen seem to treat it as a collection of tools, and causality in computer science (AI) seems like a separate subfield with the do operator and all that.
I guess you may be familiar with Judea Pearl’s work already, but he did write a popular treatment of the subject in _The Book of Why_. I’m not trying to put computer scientists on a pedestal, but there is something about the uncompromising rigor that comes with putting abstractions inside brainless machines.
Thanks for the tip, I'd read some material of his from the 1980s that helped me to start understanding bayesian networks, it inspired me to revisit my university statistics course which was fruitful. I am a sucker for pop sci!
Sometimes it's so hard it seems almost impossible. People are still debating whether there is casual relationship between lead and violence.
Or aluminum and neurotoxicity.
I agree that people could have a more effective BS filter, but I think there’s also some obligation to not publish BS in the first place at the risk of your trustworthiness reputation.
People intentionally or repeatedly inadvertently publishing BS should be called out and their opinions down-weighted heavily.
People having a good BS filter is one step in the chain towards this down-weighting, but isn’t the whole chain. If anything, people slinging BS today are likely to gain additional reach and opinion-weight, rather than to lose it.
> should be called out and their opinions down-weighted heavily
But thinking about this from a loosely signal-process-y angle, once you have a BS signal in one channel that routes to a a particular set of minds, how does one route the corrective signal to that same set of minds (as opposed to, say, just your local friends who already agree with you anyway)? And even supposing you do, those minds are already programmed to process these signals in particular ways that probably means they're predisposed to accepting one or the other before either signal even arrives.
The answer to this has always been to address the root of the problem, the preprogramming, through (relatively) uniform mass education, because that's the one and only place where you can (relatively) uniformly cram ideas into everyone's brain before they scatter to the winds and begin fancying themselves "free thinkers".
If you integrate over time, an erroneous signal (a stray bit of noise) can still result in a few minds here and there having the wrong information. That’s not ideal, but it’s unavoidable and not catastrophic so long as the source of the bad information systemically has reduced amplitude next time.
Right now, sources of bad information get systemically louder, not quieter, over time. That’s way worse.
Yup, it is called establishing a reputation, positive or negative, trustworthy or untrustworthy
This works well naturally in small communities, but once the society becomes too large, BSers, liars, and scammers can just move on to a new gullible crowd.
So we got social media with review & reputation management systems, and these are now, of course, promptly gamed to the max. Moreover, this gaming is being done by the very people who should be most de-amplified, in order to amplify their BS, inlcuding everyone from just trolls to professional RUS dezinformatsiya shops (that'll gather a lot of downvotes).
So yes, a huge part of the solution is to de-amplify the crowd that spews Bs or deliberate lies.
Sure, freedom of speech is a right, but no one has to be required to amplify you on their platform - that takes away the freedom of speech of the platform owner (e.g., if HN were required to amplify everyone, then moderation would become effectively illegal).
Why does this view gain downvotes so frequently? IDK, but it seems to be mostly readers with no nuance who think that freedom of speech requires zero restrictions, so any shadow of moderation or restriction rankles them, and they are not articulate enough to state a reason, but can still hit the down-arrow.
The post does exactly this, calling out Alex Berenson by name - but despite its best efforts, it will have very little impact on the people paying him to receive BS directly from his Substack.
I don't think your opinion is unpopular because it's wrong. It's unpopular because it's empirically impractical. People don't suffer a reputational hit for publishing BS, and saying "But they should" doesn't get us anywhere (as you yourself point out). How do you propose that we actually cause reputational hits for such people?
One way is to teach additional statistical literacy to the general public.
There are probably other proposals worth trying and I'd be interested in hearing your takes on what they are. (One that might be effective - though certainly controversial - is for the government to say that Alex Berenson is actively putting lives in danger with lies and that, if the untruth of his statements can be proven in court, he should be subject to criminal penalties, just like Elizabeth Holmes is on trial for lying about her blood tests. But that seems a lot less good for society as a whole than teaching statistical literacy.)
I don’t have a workable suggestion, but I agree that it’s not government regulation of speech. It’s going to be difficult to fix when enragement engagement is profitable for the platforms and the BS slingers.
Elizabeth Holmes is on trial for lying about things where she had a specific legal obligation to tell the truth. I’d never heard of Alex Berenson before today, but I doubt his situation is one in which he’s obligated to tell the truth. That’s okay. Sunday school preachers aren’t either and we’re cool with that. We’re good with the Santa and Easter Bunny myths.
The moment you go down that road, every academic epidemiologist and COVID expert will immediately be prosecuted because they have repeatedly made claims that were totally false and which have led directly to many deaths. For example, the claim that lockdowns work (they don't and this was known before COVID times), has led directly to people dying due to delayed medical treatment. Poverty also kills of course and their demands/misinformation has led to that too.
The same is also true of more or less the entire field of nutrition, none of which has yielded anything useful, and has resulted in governments themselves spreading misinformation:
https://sebastianrushworth.com/2021/11/27/is-saturated-fat-u...
Incidentally, Berenson has written a long rebuttal to this kind of thing which you can find here:
https://alexberenson.substack.com/p/the-english-data-on-vacc...
> _the claim that lockdowns work (they don't and this was known before COVID times)_
This is a bold claim and I'd be curious to see you back this up. By "lockdowns" - do you mean actual, genuine lockdowns / quarantines (mandatory stay-at-home orders, government-distributed emergency food packs and other essentials), or do you mean capacity restrictions etc. that get called "lockdowns" in the popular media?
Note that my straw-man proposal (which I'm not seriously endorsing) is not that it should be prosecutable to have been wrong. Plenty of startups try to build something, and it doesn't work; they don't get prosecuted like Holmes. Holmes is facing prosecution for fraud, for knowingly telling falsehoods. If your implication is that academic epidemiologists, COVID experts, and the entire field of nutrition are all _fraudulent_ as opposed to merely just going through the usual course of science - which, to be clear, I see as mostly but not entirely impossible - then yes, I think we have a rather serious problem on our hands, which we need to figure out for the survival of humanity, and I'll repeat my comment above: I'm very interested in knowing what proposals you have for solving it.
Yes. Take a look at table 1 in this WHO report from 2019 on non-pharmaceutical interventions:
https://apps.who.int/iris/bitstream/handle/10665/329438/9789...
"Quarantine of exposed individuals", "contact tracing", "border closures" and "internal travel restrictions" are "not recommended under any circumstances".
"Home quarantine of exposed individuals to reduce transmission is not recommended because there is no obvious rationale for this measure, and there would be considerable difficulties in implementing it"
And: "Border closures may be considered only by small island nations in severe pandemics and epidemics, but must be weighed against potentially serious economic consequences."
Also: "There is also a lack of evidence for the effectiveness of improved respiratory etiquette and the use of face masks in community settings during influenza epidemics and pandemics."
Finally: "This document will serve as a core component of WHO’s influenza prevention and control programme in community settings".
Consider that this was published in _2019_. Less than a year later every single position had been inverted, and epidemiologists / public health experts were telling the world that "the science" didn't just recommend all these things but outright required them. The masks issue is especially neuralgic because at the start they correctly stated that there was no evidence mask mandates would have any effect, then flipped their position overnight and asserted anyone who thought otherwise was ignoring science.
As you correctly argue, there must be never be punishments for being wrong. However this is not a case of merely being wrong. No new evidence came to light to cause these flips in position, and in fact - as predicted - lockdowns, masks and so on have had no impact on the course of the pandemic. The models that were used to justify lockdowns were never validated against reality, and even contained severe coding errors: the people behind them have never admitted they did anything wrong here. Moreover the people who pushed hardest for lockdowns consistently claimed that there was "scientific consensus" behind what they were doing, even though just a year earlier the WHO itself had published a document stating that their recommendations should never be used in any circumstances! In other words, they were lying.
And that's before we get to the fact that the WHO has a long track record of going into mad panics over diseases that are extremely mild or may not even exist at all.
Der Spiegel, 2010, "Reconstruction of a Mass Hysteria: The Swine Flu Panic of 2009",
https://www.spiegel.de/international/world/reconstruction-of...
Unherd, 2021, "Is the WHO inventing diseases" (in reference to gaming addiction disorder),
https://unherd.com/2021/11/whats-china-got-against-gaming/
Any standard that allowed people to be imprisoned for spreading misinformation that was applied fairly would quite simply result in the mass imprisonment of public health officials, advisors and experts.
That document _does_ recommend "school measures and closures" and "workplace measures and closures" for severe epidemics, and it also does in fact recommend "internal travel restrictions" once it becomes a pandemic.
Hence my question about what you mean by "lockdown." This document actually argues in favor of the sorts of things called "lockdowns" in the media, and against quarantines - which was not my personal position, but I think it definitely does not back up the claim that lockdowns (in the media sense) do not work, either.
I would also note that right beside all of the things you quote is a column "Quality of evidence," and the recommendations saying that you should avoid home quarantines and border closures are based on evidence of "very low" quality. So I don't think that you or the WHO actually have evidence saying that these things don't work. The WHO just went with their gut feelings that they were bad ideas. Now we have facts, which supersede feelings. That is an appropriate, science-based reason to change your opinions.
(Also, SARS-COV-2 is not an influenza pandemic, it's a SARS pandemic, and this document is about influenza.)
Statistics is probably more useful than Calculus for just about everyone.
Sure. “An average human has one breast and one testicle” - N.N. Taleb
Corollary: If a human has two legs, they have an above-average number of legs.
Also, most humans have more fingers than the average.
Actually, with the prelevance of polydactyly (about 1 in 1000 people have 6 or more fingers), and less accidents happening, most people might have a bellow average number of fingers in the future.
There is no way in hell the incidence of polydactyly is that high
Edit: jesus christ you were right
Conclusion: we are not in hell
?
Are you sure? One in 500 babies is born with an extra digit.
Most will have a medical prosecute to remove it
https://en.wikipedia.org/wiki/Harrison_Bergeron
The average human has more than 1 skeleton.
(because of pregnancy)
and collectors
77% of all statistics are made up on the spot
Statistics and my AI class taught me it's 100% predictology. I mean just look at Zillow. It's the modern "tea leaf" reading.
Statistics and probability are formally defined using calculus, so that doesn't really work.
Maybe basic introductory statistics is more useful than basic introductory calculus though.
> But it's a major in modern public discourse, and one for once that I'm quite willing to lay heavily at the feet of the public education system.
Should the general population be better at such things? Yes, of course.
However, in nearly every case such things reach a broader audience via mainstream media. And how many times have we seen those entries confuse correlation with causation? We've seen it so many times that it's safe to assume it's intentional. Surely, after each incident of such negligence a teacher or professor or math savvy citizen reaches out to correct them. Yet? Never a correction or retraction?? Never a spark of "we need to educate our journalists"?
Repeat something often enough and it becomes truth in the minds of the receivers. Toss in confirmation bias and echo chambers and even if your better educated the masses, the media and those "journalists" would mitigate public's understanding.
In addition, the entire advertising-based consumer economy is based on getting the potential consumer to buy the product by whatever means available.
A public education system that creates a population conditioned to believe whatever any 'authority figure' says is also a system designed to create a population ripe for exploitation by advertisers. Similarly, the ideal authoritarian state desires a population that is generally ignorant and obedient, and that's what's been created in much of the United States.
A general population that has the tools and skills needed to independently analyze the claims of government authority figures and cable TV and Internet advertisers, that's not what an elitist-authoritarian system desires.
It's very sad to see people who completely lack these tools and skills attempting to do their own well-intentioned analysis, they're so easily manipulated by dishonest actors. They know enough to distrust 'authority figures', but not enough to conduct independent evaluations of claims. Such people have been sabotaged by the educational system.
> … there seems to be some missing BS filter where people can recognize something as silly if the example is silly enough but not in the exact same logic fail for something that seems "more reasonable somehow".
Maybe combined with motivated reasoning.
Exactly. What it is that one wants to be true has a huge effect on how one interprets something.
And our justice system is largely built off of what is reasonable, yet it's undefined.
I took high school statistics. I don't think it really had the kind of effect on students that you're going for.
Statistics like any math class was just another pointless and imposed game of symbol manipulation, for most. Not something that affects how they see the world or how they process disinformation (which disables rational thinking by appeals to emotion, so rational capabilities aren't necessarily even the issue).
Humans are naturally using statistical type reasoning all the time and are very good at it. But when it comes to things like in-group out-group consensus-forming conformity mechanisms, the whole point is they _overcome rationality_ for social cohesion. To follow rationality instead of the group, you have to leave the group, which is unthinkable, so your mind prevents you from thinking it long enough to change your mind, by emotional terminations.
That's been my observation as well. Feeding people more information or giving them more tools often doesn't change their minds, it just gives them more ways to argue a position that was decided upon by the beginning. This shouldn't be a surprise; almost no one who advocates these solutions thinks that it will change their own views on [HOT TOPIC OF THE DAY], but rather that that everyone else will come to agree with their own pre-determined position.
Another solution might be convincing people to not pay attention to so many hot botton issues and not turn everything into a debate. A certain amount of detachment is probably more healthy for individuals and for the society.
The best (and only) thing I learned in high school statistics is that no matter how many times you flip a coin, each flip has an equal chance of landing on heads or tails.
A _fair_ coin.
It is important because if you flip a coin 20 times and get 20 tails, there is probably something fishy with that coin and you probably should bet tails. If it is 100 out of 100 (or even 95 out of 100), there is effectively zero chance that the coin is fair.
This can be modeled using Bayesian statistics. You start by assuming with a reasonably high probability that the coin is fair, then you revise your assumptions as you get more and more data.
The general idea is: if a coin flip lands on tails for too many times, you should probably bet on tails. You shouldn't try that in casinos: games are seriously checked for fairness (of course, the rules make it so that the house has an edge). But in an informal setting, it can get you a small advantage.
But the lesson isn't how to tell whether a coin is fair. The lesson is that in independent events such as coin flips, lottery draws, roulette or slot machine spins, rolls of the dice, etc., past outcomes don't predict future outcomes. If people understood this they would not spend hours at gaming tables and machines, because the odds (in favor of the house) are exactly the same for your 100th pull of the lever as they were for the first one, and the more you play the more likely it is that you lose, not win.
It's not based on a false understanding of math. It's based on the feeling they get (of anticipation or whatever).
It's harder to cure a gambling addiction than just giving them math facts. They don't even accept the fact they're addicts. To accept it would feel bad so they don't have the bad feeling thoughts. This emotional stuff just doesn't yield to facts.
That’s huge though.
I think we teach math wrong -- it's taught like everybody is going to go the distance with it, whereas that's a small minority.
We should teach math _literacy_ -- how to use math to understand the world we live in, e.g. some stats as you mentioned, basic finance (the glories of compound interest), sizing stuff, etc.
Is there any society of, say, 1mil+ people that 1) doesn't have a strong government that strictly controls information, and 2) has a population that doesn't have this "major [problem] in public discourse"? Has anybody ever pulled this off?
And even then, classical null hypothesis testing is fraught with problems. Best stick to comparing models that explain the effect (instead of some default linear model where basically everything is correlation) and use Bayesian statistics.
The BS filter is only obvious because we all agree that scientifically, watching a movie can't change your medical risk.
Try swapping the graph labels with:
Took vitamin D pills,
Excercised 30mins a week,
Got the vaccine.
Most would agree that all three points probably have some level of health benefit. Suddenly the cause vs correlation can be much more difficult for the average person to determine.
Do all three reduce the death risk? Do some of them just also trend with more health conscious individuals? Are any of the three just completely bogus?
I would go as far as sacrificing calculus and vectors for stats classes if it came to a showdown.
There's an absolute jungle of information we're presented with each day, and clearly many people would just straight up buy the vaccines-cause-death statistics uncritically.
There's no critical thinking without critical stats. How are the numbers made, what do they mean? Just about every field requires you to understand this, especially the social science fields where we're talking about some quite substantial issues like replication crisis. Things like economics as well, they're things everyone wants to understand but we don't give people the tools.
Somehow we have also missed out causality, when we've had the tools for a while. I reckon it's actually quite teachable though it currently feels like an advanced subject due to historical quirks.
I also encountered recently a chart indicating that all causes mortality is higher in the vaccinated than the unvaccinated. It was clear to me, but not to the casual theater, that the median vaccinated individual is much older than the unvaccinated individual, and older adults tend to die more frequently than younger adults.
Sounds like 13% of all silly correlations cause 50% of the fallacy.
global average temperature vs number of pirates:
https://www.spaghettimonster.org/wp-content/uploads/2007/05/...
Better one is the number of pool drowning vs. Nicolas Cage movie releases -
https://i.imgur.com/q54sO25.png
I believe they are excluding the pirates off the coat of Africa:
https://www.bbc.com/news/world-africa-48581197
My god. It's linear!
Sid Meier's Civilization predicted it!
Not that many people on HN seem to be interested in spurious correlations.
https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
https://www.tylervigen.com/spurious-correlations
I like the irony of suggesting that search result demonstrates "what many people on HN seem to be interested in," when talking about spurious correlations.
Yeah, you're right.
We should make a test or something...
https://news.ycombinator.com/item?id=29370245
Oh you meant to suggest whether most people on HN were interested in a book titled _Spurious Correlations_? I guess that's the bulk of those search results too, that specific book. You should have put it in caps to indicate it is a title. I'm still not sure whether what happened to one post somehow proves whether or not "many people on HN are interested in" even that book, let alone the topic in general, but I am impressed by your confidence in your research methods.
The irony of being taken seriously does not escape me.You can perhaps correlate my amazement.
Missed title opportunity: "Do Ghostbusters Cause Ghosts?"
Another Spurious Correlations [0]. The message will always bear repeating.
[0]
https://tylervigen.com/spurious-correlations
No, most of those are "actually spurious" in the sense of the two time series having nothing to do with each other. A lot of them are pretty short series as well, so it's easy for them to be correlated.
What we are looking at is Simpson's Paradox, where the true causal relationship is obscured by information that isn't obvious from the plot.
Now before you correlation != causation, there is actually a causation here that you can access with statistics.
Simpson’s paradox is when a trend or correlation is observable in each of the sub-populations, but vanishes when the data is aggregated. For example, a drug that has a strong effect on men and women when analyzed separately, but shows little effect at the population level.
This example is not Simpson’s paradox, it is simply the misuse of statistics. Statistics, being mechanical transformations of data, only have semantics within a causal model. Simply picking variables randomly and then assuming causality when the statistics behave that way is inverting the process of knowledge formation.
EDIT: Thanks for the corrections—the _real_ data that this fictional example is based on does show Simpson’s paradox, as the dependent variable (death rates) appears to show a _positive_ correlation with vaccination status when aggregating the population, but a _negative_ correlation for every age group individually.
> _This example is not Simpson’s paradox, it is simply the misuse of statistics._
If you read to the end of the post, you'll see that the author was using this correlation to prove that the mistake is identical to another claim related to COVID [1]. This COVID-related correlation doesn't seem as spurious as the Ghostbusters one, but that's because it's much harder to spot errors like this when variables aren't so "random".
[1]: "Vaccinated English adults under 60 are dying at twice the rate of unvaccinated people the same age"
Isn't that what's happening here? There's a guy who claims vaccination has the opposite effect to what people normally claim, that they actually cause people to die. And it seems to be because people from age 10 to 59 have been aggregated. The sub-populations, for instance ages 10-19, 20-29, etc would not be showing that vaccinated individuals are dying more often. It's only by aggregating them you get the wrong conclusion.
You’re right, I did read to the end of the post to see that but for some reason I fixated on the “contrived” example and the fallacy that it entailed.
It's still interesting to note that we assume "nothing to do with each other" with some cases, "obscured" relationships on other like the OP's second point.
Where it falls relies on the viewer's knowledge of the problem space, which can also be limited enough to lure them into false causations.
My point would be that a single graph showing two trends without any further info should never be taken as more information than "there is two trends". You'll still be free to decide there is true causality based on other information you believe.
For more on Simpson’s Paradox, see
https://www.forrestthewoods.com/blog/my_favorite_paradox/
, which I happened to reread this week, and which essentially serves as a spoiler for this article.
For anyone as confused as me: the graph colors swap halfway through the article.
Nassim Nicholas Taleb put out a video explanation of how Simpson's Paradox applies here, which is more direct and less tongue-in-cheek compared to the post:
https://m.youtube.com/watch?v=XVRfBhy5vGI
There's no way on earth they can actually know the movie-viewing history of those who recently died. So the data must simply be fabricated. It undermines the point they're trying to make. They should have used a real example.
Its so so clearly a joke. Come on man.
I read GP as recognising but criticising the joke - saying that (with quite a bit more thought/effort probably) you could construct one as jokey and ridiculous/unbelievable but based on real sources to make the point with actual data.
Using the real data is the nice part of many xkcd posts, for example
Or an advertisement for the new movie.
From the about page:
> Being keenly aware of political biases on both sides, my goal is to try to remain as apolitical as possible and try to filter out what I perceive as political biases and describe what I consider to be key insights gained from a particular report or resource
It’s hard to tell what’s a joke anymore.
What's the joke? It seems to me they're trying to make a point about the study that appears lower on the page, the one claiming that vaccinated people are dying at a higher rate than unvaccinated.
The joke is that not taking into account all known correlated predictors when performing a statistical analysis can produce absurd results. In this case, age is a metric variable known to be a correlated predictor of what movies an individual has seen, the probability of their death due to COVID-19, and the probability of their being vaccinated against COVID-19. The precise relationship between those four items is beside the point here -- the point is just that if you did not take into account that age is a metric variable and that it is a correlated predictor of both vaccination status and death, then you did your analysis incorrectly.
That's not a study. It's a blog post.
Older people die more often than younger people. More older people are vaccinated than younger people.
The study at the bottom shows that vaccinated people are dying at a higher rate than unvaccinated people. The joke is on the person who concludes that the increase in death rate is because of the vaccine and not the age difference.
It is a joke as the author states in a more recent article[0]
[0]
https://www.covid-datascience.com/post/what-do-uk-data-say-a...
They didn't fabricate their data.
They simply switched the labels.
Not very well. You can see they didn't change the axis labels in this image here:
https://static.wixstatic.com/media/cf58cd_449149dafb04485eb2...
So the data is valid, the point is valid, but the point isn't about ghostbusters.
What are you talking about? Of course there's a way to know their history. You think Netflix, Prime, Hulu do not have a very detailed list of activity from their sites? This person also probably did not use cash for their recent theater visits, so their ticket purchases could be found as well. If they did it through a website, then the exact movie could be found. If they did it at the ticket booth at the theater (who does this anymore?), then with some extra effort of finding the time of the transaction and then comparing to start times you could narrow down if not determine exact movie.
In the days of big data, to state that exact behaviors could not be determined is just not trying very hard.
Oh come on. Everyone else saying it's an obvious joke and now here you are claiming it's obviously a legitimate claim. Even though you can imagine a way that this information might have been collected, it obviously wasn't.
Really? This was able to be read without a tounge being firmly planted in cheek?
Sorry. Lil shell shocked at the moment by the responses.
I'd really be interested in a good source to develop a better intuitive understanding for statistics (ideally with datasets + jupyter notebooks or something). I use statistics, I can work with statistics (even fairly complicated things) but it always feels a bit "painful" and unnatural. I'd really love to develop a more natural feeling for statistics.
As just discussed on the wonderful BBC radio show more or less - available as a podcast and also published online at
https://www.bbc.co.uk/programmes/p0b6j20n
i think it's reasonable to infer that the ghosts are angry and are killing moviegoers
I understand why the author is laying the smackdown on Alex Berenson, but sound statistical reasoning was murdered within the first month of the pandemic and has been absent ever since.
Care to explain what you mean?
Not the person you are responding to, but I challenge you to find a single COVID related article making the social media rounds that demonstrates a reasonable understanding of conditional probabilities, and presents data in a clear, level-headed, unbiased fashion.
For instance, this article has a silly example of confounding variables (which ghostbusters you watched as a child is correlated to your age, and age is correlated to covid mortality).
It proceeds to present two graphs “debunking” an anti-vax conspiracy. The first graph specifically controls for age, vaccination rates, and size of population.
After preventing strong evidence to the contrary, the article then implies the conspiracy theory made the same mistake as the ghostbusters analysis. It provides an incomprehensible graph as evidence.
So, F’s in Statistics 101 for everyone.
> The first graph specifically controls for *age*, vaccination rates, and size of population.
> Age 10-59
Hmmmm....
The link to his dataset is missing. Show me the raw numbers, I want to make my own conclusions.
Tl;dr someone fell for Simpson's paradox [0] and wrote a (probably 'viral') post on how Covid-19 vaccines are killing us; this is a (slightly confusing at first without context, I thought - especially because it starts of joking about correlation/causation that seems a bit different to me) rebuttal.
[0] -
https://en.wikipedia.org/wiki/Simpson%27s_paradox#Correlatio...
It seems to me that Berenson didn't "fall" for Simpson's paradox. He's a smart person. He peddles inflammatory rhetoric to build an audience.
For example, he portrayed New Zealand's 2-week quarantine requirements for incoming travellers to New Zealand as "indefinite confinement for New Zealanders". When people pointed out the gulf between what he was saying and what the law was, he doubled down. I'd share the exact tweets, but unfortunately his account is now suspended. Twitter really needs a way to access tweets from banned accounts in a way that removes their virality but preserves them for posterity.
What's the motivation for that? Infamy?
Since it's Sunday morning and my religious tradition is on my mind, I hope you won't mind me posting a theory based in that tradition. The response to the pandemic has reminded me of nothing so much as the demons in C. S. Lewis's _The Screwtape Letters_. Their goal is not simply for people to die - if they die innocent, they haven't gained anything. Their goal is for people to die while being culpable of great evil. At one point, the trainee demon Wormwood is excited to hear that his target might be called up to fight in the front lines of World War II, but his mentor Screwtape warns him that this might not work out the way they want.
A pandemic, by itself, would be a tragic loss of innocent lives. A pandemic where half the people involved are telling each other to make themselves more likely to die, and then they all die as a result - now _that's_ a plot Screwtape would love.
Remember Satan's original plan. Are you such a sheep that you're going to be afraid of one fruit because you were told to be afraid of it? Why live in fear? Look how good the fruit is, and make use of the freedom you have to eat it. "You will not surely die."
For me, the response to the pandemic - and especially the extent to which the pandemic is much worse because of the human response to it, from so many different politicians and business leaders in so many places - just confirms that human evil is real.
PDF copy of _The Screwtape Letters_ for those who need it:
http://www.preachershelp.net/wp-content/uploads/2014/11/lewi...
I don't know his motives.
The most charitable interpretation is he believes he is seeing something that the rest of us aren't, and he is willing to nobly risk his reputation in order to warn us and, thus, protect us. He is doing us a service at great cost to himself.
That is certainly possible. I think it is more likely that he likes attention and money. This isn't an insult to him - I also like money, and sometimes, attention. For only $29.99, you may buy his new book, "Pandemia: How Coronavirus Hysteria Took Over Our Government, Rights, and Lives". The link is available in his Substack. (And presumably would have been available in his Twitter account, had Twitter not suspended it.)
It would make burning your credibility more permanent.
Deleting banned accounts whitewashes the history of the person that got banned, allowing them to repeat the cycle without changing names/pseudonyms.
That’s probably a win for engagement and trolls, but a loss for everyone else.
An audience is an audience.
I mean, from outside of New Zealand that country does look like a "reverse" open-air prison, so to speak. Granted, the policies put in place have resulted in almost no covid-related deaths, but the reality it is what is, that country is very, very difficult to come back into (or to return), and because of that most of the New Zealanders looked trapped in their own country (at leats that's how it looks to a person like myself who lives more than a half a world away).
I'm NZ born and bred. I oppose our current lockdown situation and think it should have been lifted a long time ago. I think we got extremely lucky in our first wave and I think our government handling of the delta wave (and the lack of preparation in advance of it) is shameful. I am extremely critical of our government and very disturbed by the media's handling of it. I could not be more depressed by the sorry state of our opposition.
All of that to to say that describing NZ as a "reverse" open-air prison is so far beyond absurd I don't even know what to say. For one thing, Kiwis are free to leave NZ - perhaps it's Australia you're thinking of, which really did lock its citizens at home under punishment of 5 years in prison or a $66,000 fine[0]?
0
https://en.wikipedia.org/wiki/Travel_during_the_COVID-19_pan...
Beautiful country to be “trapped” in. Lots of amazing parks and mountains and caves and beaches and volcanos everywhere. I’ve been tempted to “trap” myself there for a couple of decades now.
It’s such a prison that billionaires are voluntarily securing land and building mansions in case the whole world goes tits up. Such a terrible place. /s
https://www.theguardian.com/news/2018/feb/15/why-silicon-val...
(Note: I’m aware it’s not some perfect utopia and one of its issues is things can be expensive to live there, especially housing)
I didn't say anything about New Zealand being ugly or whatever, I did say that (from my perspective) the freedom of movement of its citizens across the border was severely limited, hence the "prison" term.
Yes, I know that (most probably) New Zealanders were perfectly free to go out to another country had they wished to do so, but once there they couldn't have come back (unless they were a multi-billionaire like the Google guy), I regard that as a prison-like system, because it majorly forces you to remain put (most probably your family, your job, your everything are located in New Zealand, you don't want to give them away). Yes, it is a system that saved lives, but nevertheless it is a system that restricted the freedom of movement of its citizens for almost two years now.
They've allowed people to leave and come back (not the entire pandemic probably, but most of it), it's just if they come back they have to stay in a quarantine hotel for two weeks, especially if they tested positive for Covid.
A quarantine hotel that a bunch of assholes were so fucking impatient that they literally climbed fences and dodged security to escape from just to grab beer or some shit and wasted millions of dollars as New Zealand tried to contact trace everyone they came in contact with and try to keep a spread on it further.
When all they had to do was chill in a hotel for two weeks and then they'd be free to do whatever.
https://thehill.com/policy/international/asia-pacific/570511...
https://www.theguardian.com/world/2021/sep/02/new-zealand-po...
https://timesofindia.indiatimes.com/world/rest-of-world/man-...
https://www.nzherald.co.nz/nz/covid-19-delta-outbreak-man-es...
https://www.rnz.co.nz/news/national/454613/covid-19-two-esca...
A couple of these might be the same person, I just did a quick google search, but I think I've seen at least a dozen of these stories out of New Zealand over the past year and a half.
That sounds to me like a reasonable criticism and discussion to have.
That wasn't the argument Berenson was putting forth. He was saying that some large number of New Zealanders would be detained indefinitely in congregate settings purpose-built by the government.
Again, I wish I could cite the specific things, but I cannot. I was left with the very strong impression that Berenson did not care about facts.
What is a "reverse" open-air prison, what part of imprisonment are you suggesting is reversed?
Not the GP, but I believe they mean that it keeps outsiders out rather than keeping insiders in.
I see. It's currently pretty darn hard for most of the world to get into the US, is the US also a "reverse open-air prison", which alarms the poster? Really, it's been almost impossible for the majority of the population of the planet to get into the USA before the pandemic. Most of the world would be thrilled if they could come if they quarantine for two weeks.
That, and the the fear (for lack of a better word) of not going out of the country because getting back in would be impossible. As far as I understand that was the feeling of many of the expats living in Singapore, had they decided to get out of the city they couldn't have come back (and hence they would have lost their lucrative job, most probably).
Alternative explanation: He didn't "fall for" anything -- he's cynically exploiting a weakness in the UK stats presentation, with the help of selective citation, on a project of self-aggrandization. Try a scroll through the blog history.
https://alexberenson.substack.com/
https://alexberenson.substack.com/p/the-english-data-on-vacc...
Berenson is one of the most prominent deniers, previously famous for his claims about marijuana being far more dangerous than commonly believed.
I'm curious how you're (as someone who it sounds like doesn't believe him/fall for it) aware of them?
I believe you, assume you're correct, he's (nor anyone else like that) just not 'on my radar' at all. I just assumed prominence/virality of at least this one post due to the existence of one (the submission) refuting it.
He is a former New York Times reporter, makes frequent media appearances, and was very active on Twitter before being suspended, so I was familiar with him before anybody had heard of COVID-19.
Nice example of 'correlation does not imply causation' and if you submit this kinda thing to me in my data analysis class I'll fail you.
Makes me cringe a little bit when global warming presentations start out with a co2 vs global temperature chart. Especially if that forms the entire basis of the analytic part of the presentation. Because I know someone (most) is just thinking "correlation != causation" and dismissing the entire thing.
But in this example we know that there indeed is a correlation between greenhouse gases and temperature. Of course you cannot restrict your argument to that, but at some point you start with some axioms.
Yes; interesting that they never bring up the chart showing we’re at the peak of a natural 125,000 year temperature cycle.
Source? The long temperature graphs I've seen look quite irregular, rather than cyclical.
Here's the smoking gun graph he's hosting:
http://donath.org/Photos/TempChange.PNG
This is what he's talking about:
https://en.wikipedia.org/wiki/Milankovitch_cycles
Here's a graph similar to his, hosted on German Wikipedia:
https://de.wikipedia.org/wiki/Milankovi%C4%87-Zyklen#/media/...
Nobody brings it up, because scientists don't believe the observed changes in temperatures are due to Milankovitch cycles, since global warming is happening on the order of decades and not millenia.
Here's NASA's take:
https://climate.nasa.gov/ask-nasa-climate/2949/why-milankovi...
Scaling the time axis sensibly will quickly reveal that the rising temperature and this cycle cannot explain the rise we see.
There are certainly overlapping effects, but the model would fail to predict temperatures.
> global warming is happening on the order of decades and not millenia
How do we know that temperatures inferred from ice-cores (or whatever) are not extremely smoothed due to natural phenomena? Can we really get resolution down to individual years such that they agree across different test sites and methods?
Well for one thing we don't have tens of thousands of years to wait for a natural ice age to counteract industrial warming that has happened within the past century or two.
But if you submit that to scientifical journal it will be published. ;)
It worked for Hitler's Mein Kampf by just changing a few words using "feminist" language, and even won some special recognition from the publishers.
> Boghossian, Lindsay, and Pluckrose wrote 20 articles that promoted deliberately absurd ideas or morally questionable acts and submitted them to various peer-reviewed journals. Although they had planned for the project to run until January 2019, the trio admitted to the hoax in October 2018 after journalists from The Wall Street Journal revealed that "Helen Wilson", the pseudonym used for their article published in Gender, Place & Culture, did not exist. By the time of the reveal, 4 of their 20 papers had been published; 3 had been accepted but not yet published; 6 had been rejected; and 7 were still under review. Included among the articles that were published were arguments that dogs engage in rape culture and that men could reduce their transphobia by anally penetrating themselves with sex toys, as well as Adolf Hitler's Mein Kampf rewritten in feminist language.[2][4] The first of these had won special recognition from the journal that published it.
https://en.wikipedia.org/wiki/Grievance_studies_affair
> Mein Kampf and intersectional feminism aren’t usually lumped together in many people’s minds, but if linked with the right language and buzzwords, left-wing academic publications apparently will accept the combination as scholarship.
https://www.washingtontimes.com/news/2018/oct/3/grievance-st...
This is a reference to this post:
https://alexberenson.substack.com/p/vaccinated-english-adult...
I had a discussion about it at work because my teammate is vocal about the alleged harm that vaccines cause and is no stranger to confirmation bias.
Shame that they lumped so many people into the same group, because if you look closer at the data for e.g. England:
https://coronavirus.data.gov.uk/details/vaccinations?areaTyp...
There's a strong correlation between age and vaccine intake, because older people were given priority.
The difference in intake among the age groups is as high as 50 percentage points.
> ... lumped so many people into the same group ...
This is the flaw in the UK aggregation, as cited from Table 4 by the blogger. 10-59 lumps nearly-invulnerable children in the same bucket with 50-something obese diabetic smokers.
Table 8, and the last line of Table 1, show properly-weighted vaccination effects.
https://www.ons.gov.uk/peoplepopulationandcommunity/birthsde...
This sort of thing is why "do your own research" is simply not plausible advice for 99% of people, even highly educated intelligent people with time on their hands.
Most of the high level problems in our civilization require years of study to even be capable of formulating a valid opinion.
The thing people have to do is to instead focus on evaluating people and deciding who probably knows who they are talking about. This is the same thing successful leaders and managers have to do when they hire. Unfortunately there is no foolproof method. Going with the consensus in a field is going to yield better than average results, but it's not a perfect rule by any means.
My take is opposite. Everybody with a bit of common sense would see such claims as problematic immediately and so they can dive deeper into the stuff. The problem is so much easier to explain if one uses the visual approach with proportional boxes for each age group. Single metric (projection) can distort a multi-dimensional data significantly, everyone knows that. But unfortunately people make the same mistake again all the time.
Agreed. No one needs to become an expert in every possible field.
What we do need to do is educate people out of new forms of innumeracy and illiteracy, where they are falling prey to basic fallacies of statistics or rhetorical tricks like motte and bailey and cat couplings.
People are generally familiar with logical fallacies like ad hominem, strawman arguments, or circular reasoning. New weapons of persuasion are getting crafted every day; we owe it to ourselves, each other, and the future humans to educate them on defense against the dark arts.
Sounds good in theory but what about those who've fallen so hopelessly down the rabbit hole? I've pointed out the statistical flaws in the tweets and Instagram posts they send me, often there's an "aha" moment. But then they'll find another article from a questionable source or tweet from some "data scientist" misrepresenting data from John's Hopkins and the process begins all over again. Debunking the same nonsense over and over again is exhausting.
I don’t have a ton of hope for some short-term (5-10 years) corrective out of the present state of affairs.
What I _do_ see happening in that timeframe is a recognition that these skills are foundational to online culture and discourse, and a push to inculcate them in forthcoming generations through education and private and public policy action.
An immune response evolving to the environmental harms of Twitter, if you will.
But yeah - that chap who keeps failing at science in your Insta is likely a lost cause.
"_But unfortunately people make the same mistake again all the time._"
Exactly what the previous comment said.
Take the "Do your own research" slogan for what it is - an attack on expertise.
It is an attempt to lower the status, and presumably power, of "the elites" (credentialed experts), mostly perpetrated by the real (monied) elites.
On the contrary, most of the people making the "do your own research" attack on expertise are not "elites" in any sense. They're the same people who say, "I could have been an X except for all the pointless math/science".
The wealthy or semi-wealthy are either riding the wave for their own benefit or don't like what the expertise says.
> evaluating people and deciding who probably knows who they are talking about
This is hard, possibly just as hard as “doing your own research”.
My explanation of the base rate fallacy is that most car crashes are caused by people who hold a valid driver's permit. The number of crashes caused by unlicensed drivers is negligible. Therefore, we should stop issuing driving licenses.
Most people I explain this to recognise the fallacy, even if they're not trained in statistics.
A huge problem with vaccine denialism is that it stems from motivated thinking. People aren't being persuaded by bullshit arguments; rather, they already agree with the conclusions, and look to these spurious correlations as emotional support.
Excellent analogy. I wonder if someone would be kind enough to shed some light on this graph:
https://www.worldometers.info/coronavirus/country/denmark/
If you scroll down to "Total Coronavirus Deaths" and click near the middle on December 11th, you'll notice a very rapid increase in deaths. December 11th was when we started vaccinating against covid.
I'm not saying that the vax is killing people, but this is a diehard argument from those who are negative against the covid vaccine.
What could a reasonable explanation be?
PS: Someone filed for more insight with our national institute of health (SSI) and they replied that following the first shots in December 2020, about 4.500 people died within 30 days. This is generally understood to be because the group we vaccinated first was already above the mean expected age, so their time was already up so to speak, ie. death could have been caused by anything, including old age.
I think it only caught the attention of the mainstream, because in all of 2020 we had a total of 1.250 "deaths with covid" whereof about 1.000 of these had more than 2 co-morbidities.
There is a huge spike in active cases right before and after Dec 11th. More cases = more deaths.
That's true, but what you can't tell from this graph was that in this period we went from 10.000 tests per day to almost 200.000, so I think that might account for must of the uptick in active cases in this period. During some of our biggest spikes the infection% has hovered around 2 - 3%.
Present day we have about 430k active cases, which is almost 10% of the population, but the % is still 2.3%.
Testing more will uncover more benign cases, that's true, but it is also possible that people are testing more due to having more symptoms.
Vaccination has continued in Denmark so if vaccines were killing people we'd expect to see that rocketing as Denmark has one of the highest vaccination rates.
I think the causality is:
More illness -> More people get tested
More illness -> More people get vaccinated
More illness -> More people die
I agree that if the vaccine was killing people, we'd be seeing mass death. Currently we are experiencing +15% all-cause mortality, where 2020 was around 0%.
The reason we are testing 200k/day is because you don't have many options for moving around without a valid test or vaccine. Ie. school, shopping, work etc requires it.
Don't you think the +15% all-cause mortality is the "mass death" you would expect to see? Isn't it interesting that Pfizer originally publically under-reported all-cause-mortality numbers in March in the vaccinated group and then went ahead and provided the correct # of deaths in the vaccinated 4 months later? It looked like by their trials all-cause-mortality was identical in vax vs unvax'ed groups, but turns out 'statistically significant' higher number of deaths in the vaccinated group. Hmmmm.
Maybe someone should explain this to Alex Berenson before he writes another SubStack post, or the thousands of people forwarding this around social media and alternative social media platforms.
Of course Alex Berenson understands this perfectly well.
He's among the utterly despicable group pushing various forms of covid denial for the benefit boosting their standing in their tribe. Whether it's for $-for-clicks or political capital or just to make themselves feel good, it's pathetic and terrible.
Speaking directly to the people doing this on HN (including in this very thread): you're going to have to go to the end of your life knowing that when things were bad, you made them worse, and got people killed who didn't have to die like that. Do you really want the biggest contribution of your life on the world to be bad one, and for petty, pathetic reasons? Really, think about it. Do you want your life to have been about fooling people about the risks of covid, covid vaccines, and the general value of science? You don't have to live a sad, pathetic, less-than-meaningless life.
If things aren't good, go out there and contribute rather than just tear things down.
Ya, right. And 99% of everyone in UK who died from COVID (of all genders and age groups) were eating bread while alive. Eating bread is a great predictor of COVID mortality!
</end_of_sarcasm>
Don't let this, ahem, gatekeep you from watching an excellent movie.
Lots of Spurious Correlations
https://www.tylervigen.com/spurious-correlations
He is referring to a statistical quantization mistake that the target, COVID lockdown critic Alex Berenson, did not make.
But he did make that mistake? Can you explain what you mean?
I'm not sure why this sort of thing impresses people, it's obvious the 1984 ghostbuster group skews old.
That's the point? It's satire
People seem to take it as some sort of impressive critique of statistics and the scientific method. Similar to "there is a Replication Crisis, vaguely speaking, so your bigot facts don't matter." No, the replication crisis is mostly a function of low sample sizes and misinterpretation, if a paper has a decent number of people and isn't being interpreted by an activist (like, ie, the social researchers themselves), it's valid.
Same thing here, this is only impressive to people who don't know how to interpret statistics.
How can you think this is a critique of statistics? People are liking it because it's a clever brand of internet snark involving knowledge of statistics. You can say it's shallow and just for people who want to flatter themselves with how knowledgable they are, but it's hardly a critique of the field of statistics, just a critique of bad statistics.
The OP article I read did not not seem to be a critique of statistics or the scientific method?
But I can tell you why the bowl of cereal that is put in front of me always correlates to it disappearing.
As Benjamin Disraeli said "There are three kinds of lies: lies, damned lies, and statistics."
Per Wikipedia [0]: “The phrase [lies, damned lies, and statistics] was popularized in the United States by Mark Twain (among others), who vaguely attributed it to the British prime minister Benjamin Disraeli. However, the phrase is not found in any of Disraeli's works and the earliest known appearances were years after his death. Several other people have been listed as originators of the quote, and it is often attributed to Twain himself.”
I had mistakenly believed that Twain originated this remark until I looked it up just now.
[0]
https://en.m.wikipedia.org/wiki/Lies,_damned_lies,_and_stati...
Why is the ghostbusters graph obviously just a photoshop relabeling of the covid vaccine graph?
I have checked the underlying dataset myself and the graph plotted above [the photoshopped one] is correct. People under 60 who watched the 1984 Ghostbusters movie are twice as likely to die as people who watched the 2021 Ghostbusters movie. The overall deaths in Britain are running well above normal.
The poster of this article is observably a liar. Perhaps there is some kind of point to be made here, but I'll wait until someone honest attempts to present it.
The first part is a joke... Keep reading and you'll get it
I ... walks away from conversation...
The introduction to the article needs more context. I don’t understand why it exists, who cares about it and who would believe that.
I explain it a few comments up.
Often, you can discover such things by reading the article.
Often, but not in this case, hence my comment.
Did you read the whole article? At the end it reveals the entire point - that failing to take into consideration the age of study participants in things like Covid death rates render false conclusions from studies.
> Did you read the whole article?
This is Hacker News! Of course not.
So a guy posted something on Facebook and got less likes than a teenage girl usually gets on Instagram. And so this another guy decides to preach to the choir because of that? Might as well write an article about microchips in the vaccines and 5G. I guess it also will hit the top on HN.
Berenson is not just some nobody. He is a prominent media personality.
https://en.wikipedia.org/wiki/Alex_Berenson
It was pretty clear to me when I read the article. Is English a second language for you? If so, then it is not surprising that you may have missed some of the meaning; as we all learned in <native language literature> class, there is a lot more to the written word than a literal word-by-word interpretation.
Wow that seemed like a lot of effort to make a point. Often twitter oversimplifies things, but that could legitimately have been a tweet of the offending graph and a few words about how older/sick people are more likely to have been vaccinated, and also more likely to die because they're old/sick.
I do like the original Ghostbusters movie, but I'm not sure that really added to the effect for me.
Changing minds with humor is better than making points in the echo chamber.
I guess you found it funnier than me, but making points seemed like a huge part of that article.
I’m not opposed to making points.
I’m commenting on the relative merit of the craft and thoughtfulness and humor that went into the attempt.
Agree to disagree I guess, but I have read a lot of things on Twitter that I found interesting and useful (amongst a lot of crap, obviously), and this I didn't like.
This feels like one of those browser addons/scripts that replaces words,e.g
"being vaccinated" -> "watched the 1984 Ghostbusters"
"unvaccinated" -> "2021 Ghostbusters movie"
Imo, this is the level of discourse we need on vaccine efficacy, risk, and data. It's funny and demonstrates some mastery of the concept of the sample bias behind it.
When 80% of the population is vaccinated, the base rate of dying of anything at all carries over with them as people change their own status from one category to the next, creating a false causality link between the base rate of dying anyway and their status change to vaccinated. I'd say now do covid and its variants, but regardless, well done.
We need smarter and more compelling people making these cases for important things instead of feeding lines to actors and poli-sci bureaucrats on television. The hardest part of the pandemic I think will have been establishments everywhere realizing that people thought they were too dumb to be believed. One of my favorite Holzer truisms is, "A lack of charisma can be fatal," and I when I look at how this has played out, it's because the people announcing information and policies couldn't be taken seriously by 30%+ of the population even when they were telling the truth - which, unfortunately, was less than the whole time as well.
Fooled by randomness.
Man's not asking the right questions!
According to the author the Ghostbuster post is a joke. Author states so in a more recent article
https://www.covid-datascience.com/post/what-do-uk-data-say-a...
Betteridge's Law of headlines says.... NO.
There. Saved you a click.
_Whoosh!_
Hey, what was that?
If the correlation says more unvaccinated people are dying, well you should go get vaccinated so you don't die!
If the correlation says more vaccinated people are dying, it’s a statistical artifact, it’s fake news, correlation != causation, you’re a charlatan for mentioning it, why are you lying???
Kind of like how increased cases are an indication that more people should get vaccinated. But if cases have gone up in areas with high vaccination rates — why are you talking about cases, obviously you should be looking at hospitalizations and deaths, you charlatan.
This has gotten old.
Similar examples in the social sciences include systematic bias in the preparation and administration of IQ tests to different groups of people (see Charles Murray's 'Bell Curve' vs. Stephen J. Gould's 'The Mismeasure of Man').
FYI, Gould's (infamous) "The Mismeasure of Man" has been widely discredited among intelligence researchers and is considered a textbook example of an anti-science, propagandistic work, even denying basic things like brain evolution, whereas Murray's _Bell Curve_ has held up pretty well in the sciences (being hated, of course on the humanities side), and his latest _Human Diversity_, is a good overview of the state of the art, with 50% of the book being footnotes. There has been a lot discovered since Gould's _Mismeasure_ was published, and all of it bad news for Gould and good news for Murray.
The insistence, by some, that evolution only works from the neck down, remains a persistent anti-scientific belief and as Jonathan Haidt points out, the main source of science-denying on the left. There are always a stream of people like Gould who try to defend this position by mocking/caricature the opposing side, but that is not considered good science.
Please don't take HN threads on generic flamewar tangents. We don't need a thread like this to veer into yet another race-IQ war.
We detached this subthread from
https://news.ycombinator.com/item?id=29369974
.
Also, your account appears to be using HN primarily for ideological battle. We ban accounts that do that, regardless of which ideology they're battling for, because it's destructive of the core value of this site (intellectual curiosity). If you wouldn't mind reviewing
https://news.ycombinator.com/newsguidelines.html
and sticking to the intended use of HN, we'd appreciate it. We've had to ask you about this before, so please fix this.
OK, fair enough, Dang!
> The insistence, by some, that evolution only works from the neck down, remains a persistent anti-scientific belief and as Jonathan Haidt points out, the main source of science-denying on the left.
Well, at least we know where you stand on that subject then. It may be a source of science denying on the left, I'm not aware of that but I am aware that it is is a main source of racism from the rabid right.
> It may be a source of science denying on the left, I'm not aware of that but I am aware that it is is a main source of racism from the rabid right.
This attitude is really not helpful. Science is never racist. It may not meet your expectations, but that's something each individual needs to come to grips with, it's not a reason to deny science, especially something touching very important policy questions -- e.g. is it true that the SAT test is racist because it highlights group differences?
That's a question that science can answer, and throwing around wild accusations of racism against everyone who tries to study this issue objectively does not promote public policy or science. That leads to a society in which key policy initiatives are based on a denial of human nature.
And here there is a special problem with the (historical) left because going back to Rousseau and certainly through Hegel and his admirers, there is a belief that human nature is fundamentally malleable and plastic. That everyone is made of the same soft clay that can be molded by education. But this is just false. And science is clear that it's false. Yet denying this basic truth of human nature leads to terrible education and social policies. And we are seeing this play out with significant human damage in California's education policy. Large numbers of lower income students are being harmed because people in California's education bureaucracy continue to deny science.
So please pay attention to what is true, rather than focusing on whether certain scientific conclusions are culturally acceptable to you.
Actually I read both books years ago and "The Mismeasure of Man" was far more sound in its analysis. In particular the biases in IQ testing were made very clear in Gould's work.
My take on what the Charles Murray crowd (and its antecedents dating back to Francis Galton's Social Darwinism) is that it's just an attempt to justify the social status quo based on 'genetic superiority of the ruling class'. This is merely a replacement for the previous (and completely discredited) religious justification for the social status quo, i.e. priests telling the serfs and slaves that 'The gods have blessed the divine kings'. We also have the economists getting in on the game, telling everyone that gross wealth disparity is the inevitable result of pure econometric theory and so on.
For other hilarious takes on this, I always point to the "Nobel Prize Sperm Bank"[1]
[1]
https://www.thecut.com/2019/08/what-ever-happened-to-the-mys...
So notice how quickly blank-slate side devolves into politics, questioning motives, casting dark aspersions, etc.
The reason is that they don't feel comfortable talking about the science, and so this is all a form of science-denial.
If you care at all about what is real, as opposed to fighting political battles, you may want to read up on the current state of the art. _Human Diversity_ is a great read, but if you are short of time, there are lots of scientific articles out there - check out the tweet stream summarizing the data with links to the current research here:
https://twitter.com/a_centrism/status/1211170458902487042
Unfortunately trying to get people to look at scientific questions using scientific methodology is as hard as trying to convince a creationist.
They are simply not going to listen and will accuse you of being a bad person, etc.
This article is like the factcheckers fact checking a similar irrelevant thing to avoid having to investigate the real issue
Sorry I am confused about which removed chart I’m supposed to see here. If you are talking about “3. Weekly mortality rates for deaths involving coronavirus (COVID-19) by vaccination status“ then it’s in no way clear to me that your conclusion is correct. First, the fully vaccinated line is the lowest there. I can definitely see a correlation between people getting their first shot, getting adventurous because they feel safe, and then catching COVID that they aren’t prepared for because they aren’t fully vaccinated. Also this chart isn’t separated by populations. That is the mortality rate isn’t “per 100,000 fully/partially/un-vaccinated”. It’s per 100,000 total. Which means that as the total number of unvaccinated people declines so does that particular mortality rate. If everyone was vaccinated and people still died from COVID then of course only the vaccinated would die, right?
What chart for October and November? Your link only goes to 24th September and says it's the latest release.
Edit: this comment and others now appear to be nonsensical because OP essentially replaced their comment with an entirely different one.
> This article is like the factcheckers fact checking a similar irrelevant thing to avoid having to investigate the real issue
No. This article is like factcheckers satirically explaining how statistics work to an author that clearly has no clue. In fact it’s not _like_ that, it _is_ that.
I'm seeing a slight upward trend in the "1 dose" line (indeed just crossing the "unvaccinated" line in a latest month), but the "2 doses" line is much lower than both, and doesn't seem to be changing much.
You're totally right, misread the chart. Edited all out unless I find the weekly report
Still why would people with 1 dose be dying more than unvaccinated?
All the learned studying the numbers, trying to figure out what number constitutes a genocide, while doing nothing about the genocide that is happening.
This poison is just that. Poison.
The immense numbers of injury and death being caused by the jab, and not by a fucking flu, is sad to watch.
You don't need to see numbers, or debate on correlation of ticker tallies, when one is seeing it.
Good luck to those who think this is not happening, you will be brutally awaken very soon, because its worse than you think.
Creatively using statistics happens on both sides. Example: "Case statistics show us that this is a pandemic of the unvaccinated!"
But due to the 3g rule (access to events or buildings is granted to the vaccinated, recovered or _people with a PCR test less than 24h ago_) in countries like Germany, guess which group is tested most: The unvaccinated. Other groups do not need tests!
Actually, the "pandemic of the unvaccinated" is also supported by their over-representation in hospital admissions and deaths.
https://lawrence-robinson.medium.com/vaxxed-vs-unvaxxed-hosp...
"A study done by the CDC in New York accounted for the following — “A total of 1,271 new COVID-19 hospitalizations (0.17 per 100,000 person-days) occurred among fully vaccinated adults, compared with 7,308 (2.03 per 100,000 person-days) among unvaccinated adults”."
Many of those statements are about the relative numbers of infected people in hospitals.
There could be value in reducing spread due to vaccinated people (depending on how much of it there is, as you say, we don't have good surveillance of it), but there's lots of value in reducing hospitalizations, and that's what many of those statements are about.
Great, but are we just gonna forget all of this when reading "data-backed" social justice activist claims?
Yes, smugly mocking people's very valid concerns will surely win them over. Don't study the figures, just do as we say!
It's not (just) smug mockery. It's taking a post that might have made a big impression on people, and giving a deliberately silly variant to make it clear how the logic is incorrect.
Just pointing out the faulty logic would've made for a boring post that many people wouldn't have bothered reading, and would risk being dismissed as hand-waving or rationalizing. Picking something eye-catching and obviously ridiculous keeps people's attention, and also can't be dismissed as mere rhetoric since the conclusion does actually hold.
Did you study the figures in the blog post? On what basis do you believe the concerns are "very valid"?
I do agree with you that mockery is unlikely to win hearts and minds, but the whole problem here is that the concerns are _not_ valid and the people with concerns are _not_ studying the figures and the people they're arguing against are _not_ saying "just do as we say," and yet the concerns persist.
That and once you start making television appearances and publishing books, your claims are fair game.
You think that watching one film over another is more likely to cause you to die?
These same people are willing to take monoclonal antibodies which are human antibodies harvested from vaccinated genetically modified rodents. It’s the same spike protein antibody just produced in rodents instead of their bodies.
The reason anti-vaxxers are being mocked is because their entire MO is pointing out the sawdust in the eyes of others while ignoring the plank in their own. Calm and reasonable debate only interests them when their own so-called "valid concerns" are being put into question, but this is thrown out the window when they themselves find something new to misunderstand and publish to mislead millions for likes and five minutes of fame.
Keep calling them anti-vaxxers. It's definitely win them over.
<soapbox>
I don't know how to explain this other than movie-caused mortality.
The above sentence alone gets my Skepticism antennae whirling, let alone the majority of the (relatively unassociated) facts.
What the heck is movie-caused mortality?
An exact search[0^] returns this same HN post, no help there.
An inexact search[1^] returns results mainly about deaths that occurred during the filming of a movie or TV show. These deaths are very much not the point the OP is trying to make.
Issue #1): the sentence quoted above is non-sensical, as the OP's purported method for "how to explain this," doesn't explain anything, let alone anything related to the point OP is trying to make.
Issue #2): Even if "movie-caused mortality" was a valid explanation... Not knowing how to explain something doesn't allow one to arrive at _any_ conclusion (other than, "the thing can not be explained," pedantically speaking).
Frankly, it makes as much (as little!) sense as, "I don't know how to explain this other than time-traveling telepathic murder bots."
</soapbox>
[0^]:
https://duckduckgo.com/?q=%22movie-caused+mortality%22
[1^]:
https://duckduckgo.com/?q=movie-caused+mortality
Might want to re-read the article. I think this one flew way over your head. Unless your comment is satire, in which case your comment flew over my head.
This morning, I heard about the OP's article on the podcast "More or Less." Hats off to OP, and shame on me for getting caught in the trollnet. :D
. .
BTW, More or Less is a great data science / stats podcast!
.jpg){kind=link}