0552: "Correlation"

[markfiend]

I wish more people knew correlation does not imply causation. Like antivaxxers

Antivaxxers don't even have correlation, never mind causation.

[gam0vr]

This comic reminds me of the thoroughly well-proven fact that the decline of 18th century style pirate is the direct cause of global warming, as illustrated by this graph:

¡This cheese is burning me!... I liked the comic, but I liked this post more... this graph makes no sense! It looks like they put the X-axis on the data labels, put the data values along the X-axis, then just made whatever curve they thought looked pretty. I don't know what's worse, the ludicrous statement they were mocking or the patently incorrect way they did so.

Um you don't get it. See as time has progressed there has been a decline of the pirate population. Since pirates have been declining the global average temperature has been rising. The only possible conclusion from this correlation is this:

Pirates are cool.

This statement has the benefit of being self-evident as well, I mean, who doesn't agree?

[brunswikstu]

definately gonna show my math teacher friend this one. Good comic.

[Penguino]

By observing Saturn you can get rich, as Saturn affects the stock markets :

http://obsessivemathsfreak.org/stuff/correlation/Saturn-SNP.html

[william]

Of course this Saturn strategy is still better than listening to CNBC.

[Beacons!]

Straight-up wit this time, no fooling with impossible pop culture references, I love it!

[phlip]

Um you don't get it. See as time has progressed there has been a decline of the pirate population.

Take a closer look at the labels along the X axis. They're not even monotonic, let alone evenly-spaced.

[sje46]

Um you don't get it. See as time has progressed there has been a decline of the pirate population.

Take a closer look at the labels along the X axis. They're not even monotonic, let alone evenly-spaced.

This is kinda better . . .

I don't understand why time is a variable in an X-Y graph. There should only be three variables. Otherwise, we need a X-axis.

[Titanium Dragon]

Oh man. That's awful.

I was getting on to say "This comic wasn't actually all that funny," then I realized why it was funny.

Well played.

[Gero]

Great joke! Anybody ever heard the story about the bigmouth frog that swallowed himself? This is like making statistics swallow itself. Too kewl.

[ckannan90]

I thought xkcd comics of recent times were a bit sub-par (wrt the high standards set by itself), but this one was classic xkcd!

[MrGuy]

It used to be the case that when the developers told me "Response times for the website dropped by 40% after you moved us from a local disk to the filer," I'd change it back as soon as I could schedule the downtime (usually midnight that night). Now I respond "Wow, that's a really interesting correlation. Come back to me when you establish definitively what the cause is."

My job satisfaction is up 80% since instituting this policy (though, of course, that could just be a coincidence).

[martin878]

Randall, you dick.
...
Not cool, not funny, not a good comic.

Still loving it. What's up with your avatar tho? Not cool, not a good avatar, etc.

[kriel]

Though, at this point, I have to ask, at what level of correlation can you assume causation?

I mean, Say I have an issue with a computer wire being loose. I plug it in, it's fixed. I unplug it, it's broken again. I plug it in yet again, and it works once more. I think anybody with a basic level of common sense would say that yes, the wire was causing the problem. However, it's really only a correlation. It could just be a huge coincidence.

Or am I missing a step in the proof somewhere?

[oddsock]

Though, at this point, I have to ask, at what level of correlation can you assume causation?

I mean, Say I have an issue with a computer wire being loose. I plug it in, it's fixed. I unplug it, it's broken again. I plug it in yet again, and it works once more. I think anybody with a basic level of common sense would say that yes, the wire was causing the problem. However, it's really only a correlation. It could just be a huge coincidence.

Or am I missing a step in the proof somewhere?

Yes. All that stuff you did.

[popprocks]

Randall, you dick.
I hate you so called "scientists" explaining away the Truth by using your faux-philosophy.
"Oh, the reason why the amount of atheists in a city increases with the amount of crime isn't because God-less evil baby-killing atheists cause more crime, but because as the population goes up, so does the amount of people in every demographic you see there is a third variable in which the cotangent of sigma x bla bla bla."
Don't give me that, obscuring everything with complicated math and logic. This is called a proof by verbosity, and it is a logical fallacy. I took Do you really mean to suggest that atheism doesn't cause crime? Come on. ATheists reject the word of Jesus Our Savior and spits in everyone's faces. Of course the more they increase in a locality, the more morality goes down. STop making us into an unloving, uncaring, unGodfearing country, with your God denying, truth twisting atheist douchebag scientists, and let the Love of Jesus Fill your heart!

Not cool, not funny, not a good comic.

Judging by that mass of facetiousness and spiritual satire, I no longer have to guess your religion.

[csours]

A true classic in its own time. I will laufgh every time I see this one.

@popprocks, I would guess republican.

[Naurgul]

This one is exactly like Schrödinger's Comic, except it replaces humour with causation.

Also...

Randall, you dick.
I hate you so called "scientists" explaining away the Truth by using your faux-philosophy.
"Oh, the reason why the amount of atheists in a city increases with the amount of crime isn't because God-less evil baby-killing atheists cause more crime, but because as the population goes up, so does the amount of people in every demographic you see there is a third variable in which the cotangent of sigma x bla bla bla."
Don't give me that, obscuring everything with complicated math and logic. This is called a proof by verbosity, and it is a logical fallacy. I took Do you really mean to suggest that atheism doesn't cause crime? Come on. ATheists reject the word of Jesus Our Savior and spits in everyone's faces. Of course the more they increase in a locality, the more morality goes down. STop making us into an unloving, uncaring, unGodfearing country, with your God denying, truth twisting atheist douchebag scientists, and let the Love of Jesus Fill your heart!

Not cool, not funny, not a good comic.

You sir deserve your very own comedy show. Your talent is wasted here.

[MrGuy]

i actually very recently got into an argument about this in a chat. turns out, the actual statistical definition of "implies" is different from the day to day usage, and it does actually mean "proves", which is what my argument was, that it does imply causation, but doesnt prove it.

There are a lot of problems with assuming all correlations are strongly suggestive of causation (in the colloquial if not necessarily the mathematical sense)

It's not always obvious which way the causation lies. Lets say we notice a correlation between falling property values and rising crime. Is it the rise in crime that drives down property values, or is it the lowering of property values that causes more crime?

It's also not always the case that either of the two correlated observations cause the other. I notice that on days when it's too hot in my apartment, I can only get cold water in the shower. Does "my apartment is hot" cause my showers to be cold? Or do cold showers make my apartment hot? Neither, actually--they are both symptoms of the same external cause (the boiler's on the fritz again).

Some correlations make no sense as causation. It was the case for a long period of time that "which conference won the Super Bowl last year?" was highly correlated with "how stocks do in the next year." Should you invest based on knowing who won the Super Bowl, on the assumption that it's causitive of stock market movement?

Assuming correlation implies causation can lead to wrong-headed solutions to a lot of problems, because you spend time trying to drive "metric correlated with what I actually want" on the hope that it will make the metric you care about go up, even when there's no clear mechanism that suggests it will help. e.g. "I've noticed that when widget sales are strong, we get more units returned as defective than we get when widget sales are weak. Widget sales are down. We need to do something to increase the number of widgets returned as defective, so we'll have better sales!" This example is obviously silly, but I've heard similar "assume driving the corrrelated statistic will drive profit!" arguments from a number of supposedly smart highly paid execs in my day.

[jc]

Straight-up wit this time, no fooling with impossible pop culture references, I love it!

Oh, I dunno; one might argue that the idea that correlation means causation is a pop culture thing. After all, technical people of all sorts, and scientists in particular, usually get educated out of such nonsensical beliefs at an early age. For at least a century, minimal statistical understanding has been a part of scientific and engineering education. Inferring a causative effect from a correlation is something done primarily by politicians and others trying to convince "the masses", so it really does qualify as a "pop culture" phenomenon.

[markkat]

One of my favorites. Well done.

[Lasher]

Gotta say that I love this comic.

[Geogriffith]

Though, at this point, I have to ask, at what level of correlation can you assume causation?

I mean, Say I have an issue with a computer wire being loose. I plug it in, it's fixed. I unplug it, it's broken again. I plug it in yet again, and it works once more. I think anybody with a basic level of common sense would say that yes, the wire was causing the problem. However, it's really only a correlation. It could just be a huge coincidence.

Or am I missing a step in the proof somewhere?

http://xkcd.com/242/

[VDOgamez]

Randall, you dick.
I hate you so called "scientists" explaining away the Truth by using your faux-philosophy.
"Oh, the reason why the amount of atheists in a city increases with the amount of crime isn't because God-less evil baby-killing atheists cause more crime, but because as the population goes up, so does the amount of people in every demographic you see there is a third variable in which the cotangent of sigma x bla bla bla."
Don't give me that, obscuring everything with complicated math and logic. This is called a proof by verbosity, and it is a logical fallacy. I took Do you really mean to suggest that atheism doesn't cause crime? Come on. ATheists reject the word of Jesus Our Savior and spits in everyone's faces. Of course the more they increase in a locality, the more morality goes down. STop making us into an unloving, uncaring, unGodfearing country, with your God denying, truth twisting atheist douchebag scientists, and let the Love of Jesus Fill your heart!

Not cool, not funny, not a good comic.

Do you know what's sad? I actually had to check to see if you were being serious or not. Some people at school actually argue JUST like that and I'm sick of it.

But that isn't nessicarily why I felt the need to check.

[zapakh]

Though, at this point, I have to ask, at what level of correlation can you assume causation?

I mean, Say I have an issue with a computer wire being loose. I plug it in, it's fixed. I unplug it, it's broken again. I plug it in yet again, and it works once more. I think anybody with a basic level of common sense would say that yes, the wire was causing the problem. However, it's really only a correlation. It could just be a huge coincidence.

Or am I missing a step in the proof somewhere?

I would personally agree that you have established causation. I could be wrong, but then again my constituent atoms could spontaneously tunnel to Mars. However, in order to establish causation, you must at some point stop talking about your observations and begin talking about what you believe about the way the system works, and how its parts interact.

Here, you invoked common sense, and this is the point at which you rose above the realm of observation and entered the realm of philosophy. The specific verbiage can take many forms, but unless I'm mistaken, a transition to philosophy is required in order to establish causation, because causality doesn't figure into the observe-hypothesize-test cycle (except perhaps as motivation for the hypothesis).

[abstractpotato]

Randall, you dick.
I hate you so called "scientists" explaining away the Truth by using your faux-philosophy.
"Oh, the reason why the amount of atheists in a city increases with the amount of crime isn't because God-less evil baby-killing atheists cause more crime, but because as the population goes up, so does the amount of people in every demographic you see there is a third variable in which the cotangent of sigma x bla bla bla."
Don't give me that, obscuring everything with complicated math and logic. This is called a proof by verbosity, and it is a logical fallacy. I took Do you really mean to suggest that atheism doesn't cause crime? Come on. ATheists reject the word of Jesus Our Savior and spits in everyone's faces. Of course the more they increase in a locality, the more morality goes down. STop making us into an unloving, uncaring, unGodfearing country, with your God denying, truth twisting atheist douchebag scientists, and let the Love of Jesus Fill your heart!

Not cool, not funny, not a good comic.

Actually, studies have shown a correlation between increased religion and increased crime. However, this is likely not cause and effect, but the other way around. As crime increases more people turn to religion for solace. =P


Anyways, I unfortunately did not get the comic until reading this forum as I wont get to stats for a year or two =(

[Zhar the Mad]

This one is fine, and all, but.....I had to read it 40 times before I got thte joke. Har, har, he gives an exaple of what he was talking about! Woopty dangle doo. I'm just...my mind is overly logical, and extremly illogical, hence my high IQ and high amout of meds I take, and I prefer the jokes that require less thinking, or at least less smarty-pants ness. Not saying I didn't like it, Randall, I'm saying I didn't like it as much as the Your scrabble joke thingy, but it is still very funny. I understand your intelligence exeeds mine by a wide margin, so you have every right to show that. It's just my opinion.

[zapakh]

This one is fine, and all, but.....I had to read it 40 times before I got thte joke. Har, har, he gives an exaple of what he was talking about! Woopty dangle doo. I'm just...my mind is overly logical, and extremly illogical, hence my high IQ and high amout of meds I take, and I prefer the jokes that require less thinking, or at least less smarty-pants ness. Not saying I didn't like it, Randall, I'm saying I didn't like it as much as the Your scrabble joke thingy, but it is still very funny. I understand your intelligence exeeds mine by a wide margin, so you have every right to show that. It's just my opinion.

I understand. I was like that on meds, too.

Disclaimer: Correlation does not imply causation.

[neoliminal]

I love that he's not wearing pants in this comic.

I totally LOL'd

But let's remember kids, there ARE times when causation and correlation do coincide.

"Sometimes a cigar really is a big penis in your mouth."

[BlueEyedGreen]

Yay for elementiary logic: Fallacy of Questionable Cause FTW.

[BlueLaughter]

Good simple comic, gave me a good laugh.

Causation -> correlation.

[Detached]

Randall, you dick.
I hate you so called "scientists" explaining away the Truth by using your faux-philosophy.
"Oh, the reason why the amount of atheists in a city increases with the amount of crime isn't because God-less evil baby-killing atheists cause more crime, but because as the population goes up, so does the amount of people in every demographic you see there is a third variable in which the cotangent of sigma x bla bla bla."
Don't give me that, obscuring everything with complicated math and logic. This is called a proof by verbosity, and it is a logical fallacy. I took Do you really mean to suggest that atheism doesn't cause crime? Come on. ATheists reject the word of Jesus Our Savior and spits in everyone's faces. Of course the more they increase in a locality, the more morality goes down. STop making us into an unloving, uncaring, unGodfearing country, with your God denying, truth twisting atheist douchebag scientists, and let the Love of Jesus Fill your heart!

Not cool, not funny, not a good comic.

... ATheists reject the word of Jesus Our Savior and spits in everyone's faces. Of course the more they increase in a locality, the more morality goes down. STop making us into an unloving...

... AT... ST

That's all I can think about when I see that post now...

[dennisw]

Um you don't get it. See as time has progressed there has been a decline of the pirate population.

Take a closer look at the labels along the X axis. They're not even monotonic, let alone evenly-spaced.

This is kinda better . . .

Spoiler:

I don't understand why time is a variable in an X-Y graph. There should only be three variables. Otherwise, we need a X-axis.

This comic reminds me of the thoroughly well-proven fact that the decline of 18th century style pirate is the direct cause of global warming, as illustrated by this graph:

Spoiler:

¡This cheese is burning me!... I liked the comic, but I liked this post more... this graph makes no sense! It looks like they put the X-axis on the data labels, put the data values along the X-axis, then just made whatever curve they thought looked pretty. I don't know what's worse, the ludicrous statement they were mocking or the patently incorrect way they did so.

This version is better in that the pirate count and temperature are shown as separate lines. The other two seem, to me, to link the two together in an effort to support causality, while my version only shows the potential for correlation.

pirates vs temperature_html_m6ba1f6b1.jpg (23.73 KiB) Viewed 6117 times

On the other hand, all this analysis ruins the fun and seriousness of the original graph. I think the whole point of the way it's presented is to be satirical.

[Max Chaplin]

It reminds me of this old strip:



Abusing correlation is fun.

[Istaro]

This whole "correlation != causation" could be the beginning of a very slippery slope...

Shouldn't that be "correlation ⇏ causation"?

I know, I'm being lame for responding to a first-page post on the second page, but I just wanted to dig up that Unicode character. Even though at this size it's kinda too small to read.

[chrisj]

Every now and then I consider XKCD extra funny- in that special, sui generis way.

You know, an XKCD that is just totally XKCD-y.

This was one of those days cheers!

[Evilagram]

Scrappy comic, until I had it explained. Wow. I did not see that coming.

Also, been randal'd. I literally said the phrase, "correlation does not mean causation", in english yesterday, something which I do not say often.

[Schreiber]

Snappy is the right word for it. This comic reminds me a little of your earlier work. Dare I say...you're regressing?

[TappingTheLine]

I didn't get it until I read the thread. Then my cheese turned to lava I laughed so hard.

However.

24. Ignoring All Anecdotal Evidence:
Example: I always get hives after eating strawberries. But without a scientifically controlled experiment, it's not reliable data. So I continue to eat strawberries everyday, since I can't tell that they cause hives.

In addition, look at this: http://en.wikipedia.org/wiki/Black_swan ... Background

Face it. Inductive reasoning never "proves" anything. And deductive reasoning doesn't function without inductive reasoning to support it. I don't think anybody here can "prove" that the Laws of Physics won't go haywire one day. That doesn't mean you should live in fear of getting thrown off the earth when gravity shuts down. Am I the only person who shouted "YES!" in between laughter after reading the alt-text?

[Random832]

Is there any way, purely within statistics, on the numbers alone, to get a definitive idea of which of two correlated factors causes the other (or they're really unrelated, or there's a common cause)? I don't think there is. Even time lag can't rule out a common cause. You have to have a mechanism. (that is, an idea of HOW one thing can imply the other)

So then why is it any business of statistics what does or doesn't imply causation?