ALL BLOG POSTS AND COMMENTS COPYRIGHT (C) 2003-2018 VOX DAY. ALL RIGHTS RESERVED. REPRODUCTION WITHOUT WRITTEN PERMISSION IS EXPRESSLY PROHIBITED.

Friday, January 11, 2019

Heuristics are not proof

And yet they remain useful when conclusive evidence is not available:
Faced with the 1854 cholera outbreak in London, John Snow had no idea what mechanism caused cholera, and his instruments could not reliably identify the contaminants in water supplies, but he noted what we would now call correlations: some water companies had more of their clients die than others, even though all of them supplied to rich and poor households alike. South of the river companies were more deadly, and they drew more contaminated water from the river rather than other sources, and filtered it less than other companies. Some neighbourhood pumps had more deaths nearby than did others. This was a geographic form of correlation (now called a Voronoi diagram) and it was on that correlational basis, without knowledge of the real mechanism, that he took the handle off the Broad Street pump, and stopped the epidemic.

That is the way we tell the story now, but Snow was a careful and clever man, and pointed out another explanation: the cholera outbreak was coming to an end anyway, as people ran away from areas where there were many deaths. The common folk who believed that correlation implied causation ran for their lives and lived to see another day.

Snow also had to cope with a major anomaly in his geographic correlational investigations. None of the brewery workers right next to the Broad Street water pump fell ill with cholera. It turned out that they received free beer, and the water for the beer was boiled so as to release the flavour of hops, thus inadvertently killing off the water-borne organisms.

Snow jumped to a conclusion because his mind was prepared to interpret associations in a particular way, intially by his doubts about the air transmission miasma theory and later by his own hypothesis of water-borne transmission. He jumped to the right conclusion, without proofs of the causal mechanism which were only available years later.
This is why you should NEVER try to dismiss any correlation with the idiot's refrain that "correlation is not causation". That is an astonishingly stupid thing to say, as it is tantamount to saying that "a clue is not a mathematical proof." Who claims that it is?

As Michael Woodley points out, there is never causation WITHOUT correlation. Which means that correlation is a necessary, but insufficient indicator of causation, it is not a synonym for it.

Labels: ,

53 Comments:

Blogger Cataline Sergius January 11, 2019 1:14 PM  

I remember that one.

The control turned out to be a well that was contaminated because of sewage line break.

Blogger Taignobias January 11, 2019 1:20 PM  

The word I use is "suggest", as in "Correlation of two connected events or trends suggests causation." The stronger the correlation and/or connection, the stronger the suggestion.

Blogger Cataline Sergius January 11, 2019 1:22 PM  

This comment has been removed by the author.

Blogger Nate73 January 11, 2019 1:22 PM  

I feel a small glimpse into Vox's displeasure with gamma posturing each time I hear somebody say this in my presence, as if it's some deep insight. How much of the present problems are due to the "war on noticing" that the media perpetuates?

Blogger Stickwick Stapers January 11, 2019 1:30 PM  

This is why you should NEVER try to dismiss any correlation with the idiot's refrain that "correlation is not causation". That is an astonishingly stupid thing to say, as it is tantamount to saying that "a clue is not a mathematical proof." Who claims that it is?

As Michael Woodley points out, there is never causation WITHOUT correlation.


If for nothing else, we can be grateful for Vox's ability to mercy-kill various forms of pop idiocy that really needed to die.

"Religion causes war." DEAD

"Extraordinary claims requires extraordinary evidence." DEAD

"Correlation isn't causation." DEAD

Blogger Longtime Lurker January 11, 2019 1:52 PM  

Correlation provides important clues where causative mechanisms might be found. Some observers note a correlation between increased volcanic activity and solar minimums.

Does that mean decreased solar radiation causes an increase in volcanic activity? The issue is not settled, but some observers speculate that weaker solar performance enables the greater penetration of cosmic rays into the earth's subsurface. These observers further hypothesize that the incoming cosmic radiation excites muons in magma, thereby increasing its viscosity, which increase the explosivity of sialic-based volcanoes. Think Krakatoa.

But is this hypothesis correct? I don't know. But this hypothesis might never have been formed absent the asserted correlation between solar minima and volcanic activity.

Blogger Mocheirge January 11, 2019 2:08 PM  

I'm rather disappointed with Taleb and his pissing match with Molyneux. Taleb spent so much time building up Fat Tony -- who uses heuristics and intuition -- that it's painful to see him rejecting national IQ (a proxy for measuring intelligence) as correlated with national wealth/development. The fremdscham hit new heights when he pulled out "4000y ago Meds made amazing stuff" to counter Molyneux's comment about modern Iraq's IQ. For another jolt of cringe, remember that he went after Mary Beard for extrapolating "black man common in ancient Britain" because of outliers.

Taleb's shtick is playing the academic iconoclast, but this time he just ended up virtue-signalling.

Blogger FUBARwest January 11, 2019 2:15 PM  

"Extraordinary claims requires extraordinary evidence." DEAD"

Did Vox dismantle this on this blog somewhere?

Blogger Mocheirge January 11, 2019 2:18 PM  

FUBARwest wrote:"Extraordinary claims requires extraordinary evidence." DEAD"

Did Vox dismantle this on this blog somewhere?


Is it necessary? It's such an asinine claim that it can't even prove itself.

Blogger FUBARwest January 11, 2019 2:19 PM  

Fair enough

Blogger VD January 11, 2019 2:20 PM  

Pop illogic and "extra proof".

Blogger FUBARwest January 11, 2019 2:20 PM  

Thank you

Blogger Barbarossa January 11, 2019 2:21 PM  

@6 I would also add that sometimes the causative mechanisms are irrelevant to one's purpose. Consider a mortgage lending company. They note the inverse correlation between credit rating and default on loans. What causes low credit scores? A whole host of reasons. Laziness. Dishonesty. Illness. Poor life choices. Sheer bad luck. Spendthrift spouse. Does the lender care? No. They note the correlation and either reject the loan application or jack up the interest rate charged to compensate for the risk taken.

Ditto for why most employers won't hire someone with a dragon tattoo on his face.

Blogger Damn Crackers January 11, 2019 2:39 PM  

@11 - "In summary: once something is proved, no extra proof is required. From a logical perspective, Sagan did much better when he remarked that absence of evidence is not evidence of absence."

Isn't this also validated by Bayes' Theorem? One corollary of the theorem if a specific occurrence happens once, there is a much better chance of it happening again.


Blogger Longtime Lurker January 11, 2019 2:44 PM  

@13: This is true. Not everyone requires scientific certainty about causative mechanisms. Combat profiling is an excellent case in point. Check out Left of Bang if you haven't already. LOB also pairs nicely with the 4GW Handbook by Lind and Thiele that CH publishes.

Blogger camcleat January 11, 2019 2:57 PM  

Maybe it predates it, but didn't the "Correlation does not equal causation" as a pop-ish meme start with that idiot Flying Spaghetti Monster guy?

If I recall correctly, he included in that essay a plot of global temp vs number of pirates and concluded "we need more pirates." He used that plot to show "correlation does not equal causation."

The whole essay was a fallacious clusterfrak, but the anti-Christian crowd went berserk over it like he was some sort of genius. He was a physics student or something, so that was "proof" he was smart.

Blogger Silly but True January 11, 2019 3:09 PM  

Sharyl Attkisson’s investigative reporting on vaccine-caused autism coverups has been astonishing.

It is one thing for doctors to say that parents don’t know anything about the subject, and they’d largely be right. But it’s quite another thing entirely for them to seek to dismiss parents’ experiences of having a perfectly healthy, active and responsive baby prior to vaccine ending up with a drooling meat lump within a short time after.

Lo and behold we now find out that CDC researchers long ago observed and accepted vaccines may cause autism.

Blogger Johnny January 11, 2019 3:18 PM  

Unfortunately the groups of smart people in our current university system are more given to using their abilities to sell their collective load of bunk, than they are given to coming up with a more enlightened understanding of the world we live in. And on an even less desirable level, many of them have come to believe their own rubbish, thus making them worse than uninformed.

Blogger Resident Moron™ January 11, 2019 3:28 PM  

@16

I was explaining to a colleague today that he must be really smart because he agrees with me.

So what possible explanations can there be for correlation? Just off the top of my head:

1. Coincidence? Another way of saying they're causally unrelated. A then B can lead you to the post hoc ergo propter hoc fallacy (of which correlation is not causation can be seen as merely a reformulation), sure. But when A is ALWAYS followed by B, every time you look, coincidence quickly fails as a credible explanation.
2. A causes B.
3. B causes A.
4. A and B are caused (or related) by C.

Are there any others?

Some form of causality, in the face of a pattern of correlation (i.e. beyond a sample size of 1 (cf also "enemy action"), is the only sensible conclusion available. isn't it?

Leaping from that to my pet theory is therefore proven!, as for example some notable jew theorists do regularly here, is not the same thing. It's the kind of thing idiots who think they are being smart, do.

Blogger Jon January 11, 2019 3:53 PM  

My favorite response to "correlation does not equal causation" is, "Then what does equal causation?" They have no idea and are just repeating platitudes without understanding these relationships.

Blogger pyrrhus January 11, 2019 3:54 PM  

There are an infinite number of theories about everything, but we choose one to test because we have perceived some kind of correlation or analogy...History says that most people in Britain during the middle ages drank a weak kind of ale rather than water, because untreated water was known to be associated with sickness. Apparently the alcohol in ale was enough to destroy many of the bacteria..but they didn't know it at the time, they just did what worked.

Blogger Daniel January 11, 2019 4:03 PM  

Yeah, nnt is off here

Blogger Zander Stander January 11, 2019 4:11 PM  

And another candidate for summary execution: There is more difference between individuals within a population than there are differences between populations. Slam dunked in "White Identity".

Blogger Zander Stander January 11, 2019 4:19 PM  

How about applying this dictum to some other extraordinary claims, such as diversity is our greatest strength, all people are equal, you can be born in the wrong body(and for bonus lulz, there can be six million more examples). Maybe these claims are now considered "ordinary" and self-evident?

Blogger Francis Parker Yockey January 11, 2019 4:30 PM  

@23
Lewontin's fallacy.

Blogger chronoblip January 11, 2019 4:37 PM  

Silly but True wrote:Lo and behold we now find out that CDC researchers long ago observed and accepted vaccines may cause autism.

Someone somewhere has said something about everything.

I wonder if the hiding of such results has something to do with society being "diverse and inclusive", because the connection between the two is highly discriminatory in nature.

Blogger Longtime Lurker January 11, 2019 5:19 PM  

"Correlation does not equal causation" is a thought-stopping cliche. Try it sometime. That shit works! /s

Blogger megabar January 11, 2019 5:58 PM  

Abusing "correlation-is-not-causation" is a form of fundamentally misunderstanding the underlying principle. This is rife today, and underpins much of the dysfunction of modern society.

For example, it is wrong to hate people of other races or cultures; it is equally wrong to hate dumb, fat, poor, or gay people. In times past, society was too harsh on these folks.

It was good and correct to stop such persecution. But it is a misunderstanding to then go beyond, and say that we _must_ celebrate and encourage all such things within our own culture.

Blogger Noah B The Savage Gardener January 11, 2019 6:37 PM  

"It was good and correct to stop such persecution."

This is another one of those assumptions that deserves to be questioned. On a macro scale, there is arguably nothing good about it.

Blogger Doktor Jeep January 11, 2019 7:22 PM  

And yet when it comes to the food supply, people are like deer, that literally step over other dead deer around the watering hole to drink from it. They all think their obesity, diabeetus, and cancers are just bad luck.

Blogger John Smith January 11, 2019 7:22 PM  

Access to computers and other control systems which might provide you with information about the way the world revolves around technology should be unlimited and total. All information should be free and accessible to all. That is why we at INTEGRATEDHACKS have come come up with a team of highly motivated and dedicated hackers to help you get access to information you are being deprived of. Our services include and are not limited to hacking of social media accounts,email accounts, tracking of phones hacking of bank cards and many more.
Have you ever been hacked? Need to recover your stolen account, Want to monitor your kids,spouse or partner, Change your school results track messages from an email or mobile number and many more, INTEGRATEDHACKS is the one for you. Hundreds of our clients have their phones, social media accounts, emails, servers, may bots and PCs hacked consistently and efficiently. Our professional hackers for hire team is highly qualified and can hack anything or device you desire without giving the target any form of notification which makes us one of the best.
   Are you having doubts about your partner ? Worry no more. All your doubts would be cleared.


★ Contact Us For Your Desired Hacking services via : integratedhacks@protonmail.com and experience cyber hacking like never before.

Blogger cyrus83 January 11, 2019 8:30 PM  

The pop illogic isn't helped by researchers who make strong correlation claims based on weak evidence. I became skeptical of social science research after reading some scholarly papers assigned for a psychology class where the authors were making a big deal out of an r-value of around .3 (not R-squared).

Blogger Solon January 11, 2019 8:41 PM  

This kind of thinking is what led me to supporting some of the ideas on /pol/.

Yes, everyone hates on them for being "rabid anti-jew Nazi LARPers." But then, look at what they propose:

They notice a correlation between Jews gaining power and influence in society, and that society shortly thereafter suffering cataclysmic degradation (porn, feminism, destructive communist ideals, etc.)

This doesn't suggest that ALL Jews are "evil," no more than it does that all whites are "evil." There's not necessarily a causation there. For example, we can look at the Roman Empire, and reason that Jews were not particularly influential in that society, and yet it suffered collapse anyway.

But it does suggest a pattern, that Jews in power FREQUENTLY push for destructive policies against their host society, frequently enough that a rational, logical person could reasonably assume a causation, and look for more evidence.

Humans evolved to look for and notice patterns, obviously: hunter-gatherer tribe notices that those who eat those red berries over there (holly berries) got sick soon after, they noticed the pattern, and in the future they avoided those berries. They didn't need to know specifically that holly berries are toxic, or HOW they were toxic, but they knew that "holly berry" = vomiting and diarrhea.

This is the basis of gambler's fallacy, this instinctual search for patterns.

Thusly, anyone who says "Jews aren't bad! They're wrongly persecuted!," I nowadays look at quite skeptically. We have thousands of years of patterns to go by, you really think other societies have been wrong these thousands of years, just because "correlation doesn't equal causation?"

Also, this pattern recognition suggests that women in power (voting) are similarly destructive. That one DOES fit the Roman Empire test: women gained more power (through the men seceding authority over their households and wives/families), and the Roman empire could no longer maintain their authority over their own holdings, and fell apart. Thus the cries for repealing the 19th amendment these days. Not automatically a causation, but we've seen what women being able to vote has done to our society today, there's definitely a correlation. We've got history on that one too: Socrates said "if you let women be your equal, they will soon be your superior." Not causation, but absolutely a correlated pattern.

How a society goes about solving these problems (women and Jews in power), is left as an academic exercise, but to say there's nothing there to consider is flat out wrong.

Blogger John rockwell January 11, 2019 9:00 PM  

Correlation makes the causation worthy of investigation.

Blogger megabar January 11, 2019 9:21 PM  

> This is another one of those assumptions that deserves to be questioned. On a macro scale, there is arguably nothing good about it.

Are you thinking this for a particular thing, or in general?

I don't think a society that persecutes people for something that they can't control (e.g. race, IQ, sickness) is a very good society.

Note that persecution is very different from non-encouragement, or even disapproval. That was, in fact, my original point.

Blogger HtF January 11, 2019 9:51 PM  

Seems almost too trivial to even put forth the effort. I dont envy you, Vox

Blogger Al K. Annossow January 11, 2019 10:10 PM  

"Correlation is not causation"
"Not all X are like that."
"Your mileage may very."

It's annoying but speech has to be tailored to the audience at hand or they will jump in, interrupt, and sidetrack the conversation. They feel a need to add in the disclaimer you and others already assume. And they give long, detailed exceptions.

Some subjects aren't even worth bringing up because others will over simplify a statement or concept by the next day because they don't remember complexities. That strips nuance or generalities of their true meaning. And then they think you said something very different from what you were trying to convey. Communication failure.

Blogger Al K. Annossow January 11, 2019 10:17 PM  

Barbarossa wrote:They note the correlation and either reject the loan application or jack up the interest rate charged to compensate for the risk taken.

The amount of the correlation needed to act or not act depends on the risk of not acting or acting. And on whether you have any better ideas about the causation - when investigating, you have to start somewhere and you can't rule out the possibility of multiple causative factors.

Blogger Barbarossa January 11, 2019 11:19 PM  

I was speaking in the broadest possible terms, but you are correct. That is why the lending company will view someone with bad credit but a steady employment history quite differently than someone with bad credit who is unemployed.

My point was simply that in this case correlation rules. The lending company doesn't care how someone's credit got so bad. They just know that someone with bad credit is more likely to default on a loan. The other causative factors merely determine whether they are willing to proceed adjusting for risk or to simply reject the application outright.

Blogger Al K. Annossow January 11, 2019 11:43 PM  

Barbarossa wrote:I was speaking in the broadest possible terms, but you are correct...
I didn't make it clear but I was agreeing with you and just expanding the idea to general rules of statistics. Your comment was a practical and believable example of the risk management. I also wanted to bring in the idea that erroneously acting is a different type of error than erroneously not acting - bad consequences vs missed opportunities. Cheers.

Blogger Dirk Manly January 12, 2019 12:57 AM  

This comment has been removed by the author.

Blogger Michael January 12, 2019 1:00 AM  

Someone recently told me "correlation doesn't imply causation".
My question: When a cue crashes into the white, causing it to roll forward, is the correlating impact (of the cue into the white), causative of the roll?

Blogger Bobiojimbo January 12, 2019 1:48 AM  

Thank you for this, VD. It really helps.

Blogger Zander Stander January 12, 2019 2:28 AM  

John Snow? Wasn't he guarding the wall at the time?

Blogger Al K. Annossow January 12, 2019 2:35 AM  

Michael wrote:is the correlating impact (of the cue into the white), causative of the roll?
Yes. But the cause is already known, so there was no need to calculate the correlation or even take note of it.

Correlation indicates a lack of randomness, like the common occurrence of cue balls near felt covered tables. Neither caused the other. One or more other things caused both. Your brain or further research must suss out the cause of the proximity of the two.

A high correlation between a pet dog and dog crap in the backyard also does not prove cause, but it's OK to jump to a conclusion.

Blogger Steve Samson January 12, 2019 4:52 AM  

Thanks.

Blogger Horn of the Mark January 12, 2019 6:47 AM  

Corellation =/= causation has gotten perverted in the hands of internet masses I've seen it deployed as virtually it's own logical phalacy, i.e., pointing out a correlation is a demonstration of inaccuracy per c=/=c. Madness. You can guess what type of topics were under review.

Blogger megabar January 12, 2019 8:48 AM  

Here's an anecdote about this phenomenon.

Every so often in computer science, a new paradigm comes along that promises to tame the untamable: schedules for software projects. These include the waterfall model, agile, prototyping, and so on. The results follow the pattern that early adopters are incredibly successful, and well above industry norms. Yet as the model is adopted by more people, the improvements evaporate, and projects come in just as late.

Why? Well, the early adopters are really good developers. Perhaps the new model helps them go faster, perhaps not, but either way, they're going to do much better than the average. As the model rolls out to others, they don't really understand the principles, and misuse them, or they simply don't have the talent firepower necessary to do the job well.

Originally, people used the correlation-is-not-causation veto card correctly, because they had the (proper) insight that too many people were misusing the concept of correlation. They understand that it is nuanced, and that you shouldn't automatically throw away correlations as worthless. Depending on the particulars, a correlation might be very strongly suggestive, or it may be almost assuredly noise.

As the notion rolls out to a wider audience, people who don't understand the nuance start using it, incorrectly.

Blogger English Tom January 12, 2019 10:53 AM  

@Zander Stander

I thought John Snow knew nothing!

Blogger JM January 12, 2019 12:45 PM  

"Correlation does not imply causation" is a true mathematical statement; i.e. "imply" is used here in a mathematical sense. But this statement is not true if "imply" is interpreted colloquially. Ed Tufte said it best: "Correlation does not imply causation but it sure is a hint!"

Blogger John Smith January 12, 2019 6:33 PM  

Access to computers and other control systems which might provide you with information about the way the world revolves around technology should be unlimited and total. All information should be free and accessible to all. That is why we at INTEGRATEDHACKS have come come up with a team of highly motivated and dedicated hackers to help you get access to information you are being deprived of. Our services include and are not limited to hacking of social media accounts,email accounts, tracking of phones hacking of bank cards and many more.
Have you ever been hacked? Need to recover your stolen account, Want to monitor your kids,spouse or partner, Change your school results track messages from an email or mobile number and many more, INTEGRATEDHACKS is the one for you. Hundreds of our clients have their phones, social media accounts, emails, servers, may bots and PCs hacked consistently and efficiently. Our professional hackers for hire team is highly qualified and can hack anything or device you desire without giving the target any form of notification which makes us one of the best.
Are you having doubts about your partner ? Worry no more. All your doubts would be cleared.


★ Contact Us For Your Desired Hacking services via : integratedhacks@protonmail.com and experience cyber hacking like never before.

Blogger mike January 13, 2019 2:45 PM  

But it's still good to remember that correlation isn't proof of causation. All sjws do is relay on statistics and statistics can't prove a jack it can only help you find the real cause by elimination of impossible ones.

Blogger Dirk Manly January 13, 2019 6:00 PM  

@49

"I thought John Snow knew nothing!"

You're thinking of Sergeant Schultz

Post a Comment

Rules of the blog
Please do not comment as "Anonymous". Comments by "Anonymous" will be spammed.

<< Home

Newer Posts Older Posts