Previously on Lobsters, we explored the idea that causation and correlation are distinct. Here, the distinctness goes in the opposite direction from the typical slogan, but it’s still valid: Causation does not imply correlation!
First of all, I find the point on the author writing about their own research vs other people’s research to be quite interesting.
It sounds like you’d actually recommend the book. Unfortunately, this sentiment was buried in the post
The simplest answer to this question is that the book really is wonderful, it just has this one little mistake. Noise is indeed an important subject, and three authors who don’t understand correlation and causation can still write an excellent book on the topic.
So… I feel like I don’t understand what is meant by “causation”. I feel like I use it in the sense of “force causes acceleration” or “speed of light causes increase in energy to convert into mass.” Reading your post, I feel like there’s another way of looking at it.
For example, “You can also get causation without correlation from a non-monotonic relationship.” - I would love to understand what that means.
I’m hesitant to recommend paying for dead trees, but it doesn’t sound like a bad book to have on a shelf.
I think of causality using physics. If we have two events X and Y, and all observers agree that X happened before Y, then X caused Y; X is one of Y’s many causes. The goal of many scientists is to look for statistical evidence which can point them towards possible causes for observed phenomena, but such statistics can only point vaguely.
I like to think of causation as if you forced the causes to happen, would you still get the effects? Like we might say that a thermometer correlates with the temperature in a room, but you can intervene to make a thermometer show another temperature and it wouldn’t change the temperature of the room.
By contrast if you intervened and made a room colder, the thermometer would change. In the sense we can say the temperature of the room causes the thermometer to change.
If the concept of System 1 and System 2 is valid, then we do need to come up with an explanation for the times when humans seem to put no consideration into what they’re saying. Sometimes people will utter words not just out of habitual response (System 1), but out of a non-emotional associative compulsion. In the original book, these are all treated as System 1, as fast emotional triggers, rather than as possible signs of memetic infection.
This is important partially because it gives us an explanation for how folks can be carefully reasoning using what they think are cold hard System 2 tools, yet actually be completely bullshitting themselves and their audience. It also helps explain why some people get trapped in psuedo-logical worlds where they can only appeal to authority, to memes which they’ve heard a lot, and to basic aphorisms which they feel are commonsense.
On one hand, yes, it’s a joke; the original book didn’t have a System 3. On the other hand, yes, it’s a joke; the original book doesn’t even know how many of its pages were written while its authors were using System 3 to think about their words.
The underlying critique of psychology is potent. Since causation and correlation are distinct, we require a specific and explainable mechanism of action for causes; since mental states are generally unobservable, this puts a hard barrier between neuroscience and psychology, where the latter is pseudoscientific.
Psychology studies a complex system without access to its innards. That may make it easier to get away with pseudoscience, but it certainly does not mean all psychological research is pseudoscience. In particular, if you don’t know how something works, you can nonetheless study how it behaves and responds: that’s valuable research producing valuable knowledge.
As a concrete example: the psychoeducation and psychotherapy and coaching I received as part of my ADHD diagnosis-and-treatment (a) has helped me a lot; and (b) omitted past advice that had been found to ineffective or even wrong. This was only possible thanks to a ton of psychological research (including clinical trials).
Our understanding of the human mind may never be complete, but neither are we incapable of knowing things about it; and we can and do discovering new knowledge through research.
My read: Most of the post is serious, but the “system 3” bit is a self-referential joke; note that the intro sentence “By now, we’re all familiar with the three modes of thought” is very similar to the example of “It’s an interesting formulation when someone says, “We must remember X,” where X is a false statement.”. It’s a bit like an intentional instance of Muphry’s law.
Although I probably won’t read this new pop science book, dismissing Nobel Prize winner Daniel Kahneman as “a psychologist” with a specious example about cars and hills seems like a very weird approach.
The book “Noise” contains the wrong statement “In short, wherever there is causality, there is correlation.”
This statement is wrong; to anybody versed in statistics, it is obviously wrong.
In Andrew Gelman’s blog post, the section you refer to runs like this:
The question is, how could the authors of this book have made such a clear mistake?
[…]
The authors of this new book are a psychologist, a law professor, and some dude who describes himself as “a professor, writer and keynote speaker specializing in the quality of strategic thinking and the design of decision processes.” Between them, there’s no reason to think they’d have any particular expertise in correlation, causation, or statistics.
So the point of mentioning Kahneman is a psychologist is not out-of-hand dismissal, but to point out that the mistake probably arose from lack of expertise in statistics. Psychology is one of many academic fields for which you don’t have to know the ins and outs of how correlation behaves. It could just as well say ‘chemist’ or ‘computer scientist’.
As for the example about cars and hills: it is a simple example to show how “wherever there is causality, there is correlation” is wrong. Let me quote the example again.
Imagine driving a car, reaching a hill and pumping the gas as you begin to go up so that your speed is constant. The correlation between pressing on the gas and the speed of the car is zero but they’re obviously causally related, it’s that the agent is optimizing speed!
–Example written by Rachael Meager
But the example makes no sense! The correlation to pay attention to isn’t between gas and speed, it’s between a driver who hits the gas on the hill and a driver who doesn’t. Would you describe the effects of hitting the brakes as “negative correlation”? Only if you were making a narrow argument about a single variable. The original statement is saying something more like “causes have effects” which seems trivially true. This blog post author is boring a needle-sized hole of an argument and generalizing from it to an ad hominem dismissal of at least two serious academics, both of whom I am dead-certain understand statistics very well.
You might prefer the 3 postscripts to the main blog post – they’re a bit more clear on what Gelman does and doesn’t object to. It’s not Kahneman; and it’s not Noise per se; but apparently Cass Sunstein has a history of making claims that sound good but aren’t true, and the blog post mainly seems to be a warning to read the book critically.
(Plus, the “We must remember […] causation does imply correlation” thing really is completely wrong, and Gelman is a professor of statistics, so that fish jumped straight into Gelman’s wheelhouse. If you still feel the car/gas/speed/hill example doesn’t work, Gelman names other things that that can cause zero correlation when observing causally-related variables.)
Past posts on incorrect claims by Sunstein, for you to judge on their merits:
The example only makes no sense if you take into account context that the author did not supply. In general, any control system trying to maintain a setpoint in spite of noise will exhibit no correlation between its output and the controlled variable. This seems like an egregious oversight to not point out in a book literally called “noise”.
Previously on Lobsters, we explored the idea that causation and correlation are distinct. Here, the distinctness goes in the opposite direction from the typical slogan, but it’s still valid: Causation does not imply correlation!
Oh hey, so glad that you’ve posted this here.
First of all, I find the point on the author writing about their own research vs other people’s research to be quite interesting.
It sounds like you’d actually recommend the book. Unfortunately, this sentiment was buried in the post
So… I feel like I don’t understand what is meant by “causation”. I feel like I use it in the sense of “force causes acceleration” or “speed of light causes increase in energy to convert into mass.” Reading your post, I feel like there’s another way of looking at it.
For example, “You can also get causation without correlation from a non-monotonic relationship.” - I would love to understand what that means.
Thanks, again, for the post and sharing it here.
I’m hesitant to recommend paying for dead trees, but it doesn’t sound like a bad book to have on a shelf.
I think of causality using physics. If we have two events X and Y, and all observers agree that X happened before Y, then X caused Y; X is one of Y’s many causes. The goal of many scientists is to look for statistical evidence which can point them towards possible causes for observed phenomena, but such statistics can only point vaguely.
I like to think of causation as if you forced the causes to happen, would you still get the effects? Like we might say that a thermometer correlates with the temperature in a room, but you can intervene to make a thermometer show another temperature and it wouldn’t change the temperature of the room.
By contrast if you intervened and made a room colder, the thermometer would change. In the sense we can say the temperature of the room causes the thermometer to change.
[Comment removed by author]
https://michaelnielsen.org/ddi/if-correlation-doesnt-imply-causation-then-what-does/
System 3??? I don’t remember a System 3 in the book. The linked Wikipedia article doesn’t mention a System 3.
I skimmed, saw the post was long, and came here to ask:
@Corbin, is this a joke?
If the concept of System 1 and System 2 is valid, then we do need to come up with an explanation for the times when humans seem to put no consideration into what they’re saying. Sometimes people will utter words not just out of habitual response (System 1), but out of a non-emotional associative compulsion. In the original book, these are all treated as System 1, as fast emotional triggers, rather than as possible signs of memetic infection.
This is important partially because it gives us an explanation for how folks can be carefully reasoning using what they think are cold hard System 2 tools, yet actually be completely bullshitting themselves and their audience. It also helps explain why some people get trapped in psuedo-logical worlds where they can only appeal to authority, to memes which they’ve heard a lot, and to basic aphorisms which they feel are commonsense.
On one hand, yes, it’s a joke; the original book didn’t have a System 3. On the other hand, yes, it’s a joke; the original book doesn’t even know how many of its pages were written while its authors were using System 3 to think about their words.
The underlying critique of psychology is potent. Since causation and correlation are distinct, we require a specific and explainable mechanism of action for causes; since mental states are generally unobservable, this puts a hard barrier between neuroscience and psychology, where the latter is pseudoscientific.
Psychology studies a complex system without access to its innards. That may make it easier to get away with pseudoscience, but it certainly does not mean all psychological research is pseudoscience. In particular, if you don’t know how something works, you can nonetheless study how it behaves and responds: that’s valuable research producing valuable knowledge.
As a concrete example: the psychoeducation and psychotherapy and coaching I received as part of my ADHD diagnosis-and-treatment (a) has helped me a lot; and (b) omitted past advice that had been found to ineffective or even wrong. This was only possible thanks to a ton of psychological research (including clinical trials).
Our understanding of the human mind may never be complete, but neither are we incapable of knowing things about it; and we can and do discovering new knowledge through research.
My read: Most of the post is serious, but the “system 3” bit is a self-referential joke; note that the intro sentence “By now, we’re all familiar with the three modes of thought” is very similar to the example of “It’s an interesting formulation when someone says, “We must remember X,” where X is a false statement.”. It’s a bit like an intentional instance of Muphry’s law.
On a related note Daniel Kahneman was interviewed by Kara Swisher on the Sway podcast about this new book ‘Noise’: https://www.nytimes.com/2021/05/17/opinion/sway-kara-swisher-daniel-kahneman.html (there’s a transcript as well)
Although I probably won’t read this new pop science book, dismissing Nobel Prize winner Daniel Kahneman as “a psychologist” with a specious example about cars and hills seems like a very weird approach.
The book “Noise” contains the wrong statement “In short, wherever there is causality, there is correlation.”
This statement is wrong; to anybody versed in statistics, it is obviously wrong.
In Andrew Gelman’s blog post, the section you refer to runs like this:
So the point of mentioning Kahneman is a psychologist is not out-of-hand dismissal, but to point out that the mistake probably arose from lack of expertise in statistics. Psychology is one of many academic fields for which you don’t have to know the ins and outs of how correlation behaves. It could just as well say ‘chemist’ or ‘computer scientist’.
As for the example about cars and hills: it is a simple example to show how “wherever there is causality, there is correlation” is wrong. Let me quote the example again.
But the example makes no sense! The correlation to pay attention to isn’t between gas and speed, it’s between a driver who hits the gas on the hill and a driver who doesn’t. Would you describe the effects of hitting the brakes as “negative correlation”? Only if you were making a narrow argument about a single variable. The original statement is saying something more like “causes have effects” which seems trivially true. This blog post author is boring a needle-sized hole of an argument and generalizing from it to an ad hominem dismissal of at least two serious academics, both of whom I am dead-certain understand statistics very well.
You might prefer the 3 postscripts to the main blog post – they’re a bit more clear on what Gelman does and doesn’t object to. It’s not Kahneman; and it’s not Noise per se; but apparently Cass Sunstein has a history of making claims that sound good but aren’t true, and the blog post mainly seems to be a warning to read the book critically.
(Plus, the “We must remember […] causation does imply correlation” thing really is completely wrong, and Gelman is a professor of statistics, so that fish jumped straight into Gelman’s wheelhouse. If you still feel the car/gas/speed/hill example doesn’t work, Gelman names other things that that can cause zero correlation when observing causally-related variables.)
Past posts on incorrect claims by Sunstein, for you to judge on their merits:
This thread is rapidly heading off-topic, so don’t feel you owe me a reply, although you are welcome to.
The example only makes no sense if you take into account context that the author did not supply. In general, any control system trying to maintain a setpoint in spite of noise will exhibit no correlation between its output and the controlled variable. This seems like an egregious oversight to not point out in a book literally called “noise”.
According to his Wikipedia page, he’s a psychologist:
https://en.wikipedia.org/wiki/Daniel_Kahneman#Education_and_early_career
And call me a snob but the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel is a junior Nobel Prize at best.