Do you have adblock enabled?
 
If you can read this, either the style sheet didn't load or you have an older browser that doesn't support style sheets. Try clearing your browser cache and refreshing the page.

(Wordpress)   "'Correlation is not the same as causation' has become an incantation parroted by Fox-Watchers, as part of the Murdochian campaign to undermine science and claim that nothing can ever be proved." Yeah, well, you know, that's just, like, your opinion, man   ( davidbrin.wordpress.com) divider line
    More: Interesting, Incantation, strong correlation, Gasoline, Lead, preliminary precautions, leaded gasoline, Tobacco smoking, Correlation and dependence  
•       •       •

4505 clicks; posted to Main » on 02 Jan 2018 at 3:01 AM (2 weeks ago)   |   Favorite    |   share:  Share on Twitter share via Email Share on Facebook   more»



113 Comments     (+0 »)
 
View Voting Results: Smartest and Funniest


Oldest | « | 1 | 2 | 3 | » | Newest | Show all

 
2018-01-01 08:55:06 PM  
FTFA:

davidbrin.files.wordpress.comView Full Size

"Those who use "Correlation is not the same as causation" as a magic incantation to dismiss all fact-using professions are fools holding a lit match in one hand and an open gas can in the other, screaming "one has nothing to do with the other!"

And that might lead to a literal conflagration. Because when science (reason) fails, there is only force and violence.
 
2018-01-01 09:02:59 PM  
Speaking of correlation.  
When do we address this?
img.fark.netView Full Size

fightful.comView Full Size
 
2018-01-01 09:33:03 PM  
imgs.xkcd.comView Full Size

Correlation doesn't imply causation, but it does waggle its eyebrows suggestively and gesture furtively while mouthing 'look over there'.
 
2018-01-01 09:33:28 PM  
O'course, what Brin is dodging is that it's, you know, actually true.

Because the other end of the "it turned out to be causative" rainbow is the one the anti-vaxxers inhabit, where "My daughter got a shot...and died a year later!" is a "proof".
 
2018-01-01 10:19:19 PM  

flondrix: Correlation doesn't imply causation, but it does waggle its eyebrows suggestively and gesture furtively while mouthing 'look over there'.


I've never liked that phrasing. "Correlation isn't causation" fine but the idea it doesn't imply causation is a bit much. Imply isn't exactly a strong word and you'd certainly be far more likely to find the cause of something by looking at something correlated with it than something not correlated with it.
 
2018-01-01 11:07:04 PM  

itsdan: flondrix: Correlation doesn't imply causation, but it does waggle its eyebrows suggestively and gesture furtively while mouthing 'look over there'.

I've never liked that phrasing. "Correlation isn't causation" fine but the idea it doesn't imply causation is a bit much. Imply isn't exactly a strong word and you'd certainly be far more likely to find the cause of something by looking at something correlated with it than something not correlated with it.


So you're saying Correlation and causation are strongly correlated?
 
2018-01-02 12:26:00 AM  
In other news, trumpers think they know the meanings of a couple of five-dollar words.
 
2018-01-02 02:27:45 AM  
Words with a C this week... Cool beans!
Correlation may not be, but collusion sure as fark is causation.  Continued guilty pleas (convictions) from the conspiracy of collaborators does however add up to a crap-ton of correlation, you child molester supporting, wannabe edgy-ass alt-right conservative FoxNews watching calumniators.
Ciao.
 
2018-01-02 03:10:32 AM  
Researches should stop (or at least reduce) conducting correlation tests and use more robust tests that better indicate causation. But correlation tests are easier and stats are hard.
 
2018-01-02 03:13:26 AM  

LiberalConservative: Researches should stop (or at least reduce) conducting correlation tests and use more robust tests that better indicate causation. But correlation tests are easier and stats are hard.


What you have typed means literally nothing. Have you even ever poked your head in a high school science class?
 
2018-01-02 03:13:46 AM  
"How to Lie With Statistics" was a prophetic f*cking book....
 
2018-01-02 03:14:48 AM  
Trying to read through that, it felt like the author was trying to hit "triple word score" every time they used an 8+ letter word.

Do you get paid more when your "paid per word article" uses those?
 
2018-01-02 03:21:37 AM  
"Correlation
causation
scientists
reputation
demolishing
incantation
Murdochian
campaign
matter at-hand
implications
behooved
preliminary
precautions
demonstrated"


All in the first 3 paragraphs, and that's leaving out the redundant "causitives" and "correlations".
 
2018-01-02 03:24:39 AM  
The theory of carbon dioxide as a greenhouse gas has been theorized and tested in lab scale fashion for over a century. The reverse applies here, causation requires correlation.
 
2018-01-02 03:25:29 AM  

starsrift: LiberalConservative: Researches should stop (or at least reduce) conducting correlation tests and use more robust tests that better indicate causation. But correlation tests are easier and stats are hard.

What you have typed means literally nothing. Have you even ever poked your head in a high school science class?


I have taught statistics to undergrads, I have a science phd by research, and have worked has a statistician. A large proportion of statistics I see performed in science are incorrect, generally because most researchers are not statisticians. If the initial post doesn't mean much then you may be drunk (or something). Go read the basics or correlation coefficients.
 
2018-01-02 03:25:31 AM  

pxlboy: In other news, trumpers think they know the meanings of a couple of five-dollar words.


I can say Fark and U and not spend a dime.
 
2018-01-02 03:28:20 AM  
So much for all religious miracles
 
2018-01-02 03:30:00 AM  
I get it now, The Trump administration is Dada preformance art, intended to debase the belief that government is meaningful.
 
2018-01-02 03:31:51 AM  

LiberalConservative: starsrift: LiberalConservative: Researches should stop (or at least reduce) conducting correlation tests and use more robust tests that better indicate causation. But correlation tests are easier and stats are hard.

What you have typed means literally nothing. Have you even ever poked your head in a high school science class?

I have taught statistics to undergrads, I have a science phd by research, and have worked has a statistician. A large proportion of statistics I see performed in science are incorrect, generally because most researchers are not statisticians. If the initial post doesn't mean much then you may be drunk (or something). Go read the basics or correlation coefficients.


If I went back to my university and told my physics prof he needed to do less "correlation tests", he'd turn into quizzicaldog.
 
2018-01-02 03:32:44 AM  

Harlee: And that might lead to a literal conflagration. Because when science (reason) fails, there is only force and violence.


And when force is gone, there's always Mom.

/hi mom!
 
2018-01-02 03:46:32 AM  

starsrift: LiberalConservative: starsrift: LiberalConservative: Researches should stop (or at least reduce) conducting correlation tests and use more robust tests that better indicate causation. But correlation tests are easier and stats are hard.

What you have typed means literally nothing. Have you even ever poked your head in a high school science class?

I have taught statistics to undergrads, I have a science phd by research, and have worked has a statistician. A large proportion of statistics I see performed in science are incorrect, generally because most researchers are not statisticians. If the initial post doesn't mean much then you may be drunk (or something). Go read the basics or correlation coefficients.

If I went back to my university and told my physics prof he needed to do less "correlation tests", he'd turn into quizzicaldog.


Probably. Of the ~30 postgrads I studied with in a science lab, only 1 or 2 actually knew what they were doing with stats. Yet almost all graduated, and most continued in science (and even supervised postgrad students) though their stats knowledge was poor. Correlation is best used data exploration to identify possible relationships. Once the possible relationships of interest are identified, more robust and causation stats should then be used. If research stops at correlation then it is incomplete.
 
2018-01-02 03:56:05 AM  
Fake science news.
 
2018-01-02 03:56:41 AM  
Just ask a Fox News viewer for a few examples of where it does not.  Say I'm not sure, could you explain it to me.  Follow that up with I'm sorry, but it doesn't seem as if you understand it either.
 
2018-01-02 04:10:10 AM  
So we've gone from "There was no collusion."
....to "Collusion doesn't mean anything."
....and now we're at "Nothing means anything."
 
2018-01-02 04:14:14 AM  
I thought the most far-fetched part of "V" was the "scientists-as-jews" part of the allegory.

Now, I'm finding this more and more prescient.
 
2018-01-02 04:21:07 AM  

LiberalConservative: starsrift: LiberalConservative: starsrift: LiberalConservative: Researches should stop (or at least reduce) conducting correlation tests and use more robust tests that better indicate causation. But correlation tests are easier and stats are hard.

What you have typed means literally nothing. Have you even ever poked your head in a high school science class?

I have taught statistics to undergrads, I have a science phd by research, and have worked has a statistician. A large proportion of statistics I see performed in science are incorrect, generally because most researchers are not statisticians. If the initial post doesn't mean much then you may be drunk (or something). Go read the basics or correlation coefficients.

If I went back to my university and told my physics prof he needed to do less "correlation tests", he'd turn into quizzicaldog.

Probably. Of the ~30 postgrads I studied with in a science lab, only 1 or 2 actually knew what they were doing with stats. Yet almost all graduated, and most continued in science (and even supervised postgrad students) though their stats knowledge was poor. Correlation is best used data exploration to identify possible relationships. Once the possible relationships of interest are identified, more robust and causation stats should then be used. If research stops at correlation then it is incomplete.


Your use of the expression 'causation stats' suggests to me that starsrift has the advantage of you in this argument.
 
2018-01-02 04:25:50 AM  

Mister Buttons: "Correlation
causation
scientists
reputation
demolishing
incantation
Murdochian
campaign
matter at-hand
implications
behooved
preliminary
precautions
demonstrated"


All in the first 3 paragraphs, and that's leaving out the redundant "causitives" and "correlations".


Awww. Do big words make your little head hurt? I know this is Fark, but sometimes people use more than two syllables when they speak.
 
2018-01-02 04:27:48 AM  
If only Obama have pushed Rush Limbaugh off the fiscal cliff.  Benghazi never would have happened.
 
2018-01-02 04:38:50 AM  

Dave2042: LiberalConservative: starsrift: LiberalConservative: starsrift: LiberalConservative: Researches should stop (or at least reduce) conducting correlation tests and use more robust tests that better indicate causation. But correlation tests are easier and stats are hard.

What you have typed means literally nothing. Have you even ever poked your head in a high school science class?

I have taught statistics to undergrads, I have a science phd by research, and have worked has a statistician. A large proportion of statistics I see performed in science are incorrect, generally because most researchers are not statisticians. If the initial post doesn't mean much then you may be drunk (or something). Go read the basics or correlation coefficients.

If I went back to my university and told my physics prof he needed to do less "correlation tests", he'd turn into quizzicaldog.

Probably. Of the ~30 postgrads I studied with in a science lab, only 1 or 2 actually knew what they were doing with stats. Yet almost all graduated, and most continued in science (and even supervised postgrad students) though their stats knowledge was poor. Correlation is best used data exploration to identify possible relationships. Once the possible relationships of interest are identified, more robust and causation stats should then be used. If research stops at correlation then it is incomplete.

Your use of the expression 'causation stats' suggests to me that starsrift has the advantage of you in this argument.


Up to you to decide (shrug). Am playing 7 Days to Die multiplayer (no pausing) so was being quick, hence typos etc. Regardless, correlational studies are poor, my original statement still stands.
 
2018-01-02 04:40:12 AM  
Climate change is quite real and there's more than enough evidence now to conclude human activity is not just a major contributing factor, it is 'the' major contributing factor.

But guess what? The derpy people on the right who want to say science lies to them have a slightly less derpy cousin on the left that wants to hijack the science for personal political agenda. Why yes, I do remember the Kyoto Protocol that the left huffed and puffed over because the United States wouldn't sign up for something that would accomplish absolutely nothing in terms of climate change but would do a great job screwing the United States.

And yes, that's what comes first for the left, not the climate, but sticking it the west and the United States in particular. So here's a plan.

The left gives up on 'screw the US first, climate can come after that' plans and we come up with some good ideas and the right gives up on 'science is all a conspiracy' while we're at it.
 
2018-01-02 04:42:24 AM  

wildcardjack: The theory of carbon dioxide as a greenhouse gas has been theorized and tested in lab scale fashion for over a century. The reverse applies here, causation requires correlation.


Have you heard about an IRGA (Link) it measures minute traces of CO2 using infrared absorption. This is not new science. Further, 50% of the energy we receive from the sun is in the infrared wavelengths. Just because we can't see it doesn't mean it is fake.
img.fark.netView Full Size
 
2018-01-02 04:43:00 AM  
img.fark.netView Full Size
 
2018-01-02 04:45:42 AM  

randomjsa: Climate change is quite real and there's more than enough evidence now to conclude human activity is not just a major contributing factor, it is 'the' major contributing factor.

But guess what? The derpy people on the right who want to say science lies to them have a slightly less derpy cousin on the left that wants to hijack the science for personal political agenda. Why yes, I do remember the Kyoto Protocol that the left huffed and puffed over because the United States wouldn't sign up for something that would accomplish absolutely nothing in terms of climate change but would do a great job screwing the United States.

And yes, that's what comes first for the left, not the climate, but sticking it the west and the United States in particular. So here's a plan.

The left gives up on 'screw the US first, climate can come after that' plans and we come up with some good ideas and the right gives up on 'science is all a conspiracy' while we're at it.


Most of what you said is nonsensical.
Man-made climate change exists.
Democrats (or "the left" as you say) do not want to "screw the US". That's stupid.
 
2018-01-02 04:53:05 AM  

LiberalConservative: starsrift: LiberalConservative: starsrift: LiberalConservative: Researches should stop (or at least reduce) conducting correlation tests and use more robust tests that better indicate causation. But correlation tests are easier and stats are hard.

What you have typed means literally nothing. Have you even ever poked your head in a high school science class?

I have taught statistics to undergrads, I have a science phd by research, and have worked has a statistician. A large proportion of statistics I see performed in science are incorrect, generally because most researchers are not statisticians. If the initial post doesn't mean much then you may be drunk (or something). Go read the basics or correlation coefficients.

If I went back to my university and told my physics prof he needed to do less "correlation tests", he'd turn into quizzicaldog.

Probably. Of the ~30 postgrads I studied with in a science lab, only 1 or 2 actually knew what they were doing with stats. Yet almost all graduated, and most continued in science (and even supervised postgrad students) though their stats knowledge was poor. Correlation is best used data exploration to identify possible relationships. Once the possible relationships of interest are identified, more robust and causation stats should then be used. If research stops at correlation then it is incomplete.


Science must predict, else it is useless.
 
2018-01-02 04:58:31 AM  
They're generally not that advanced here.  They usually just say "CNN fake news...
 CNN fake news... CNN fake news..." like zombies looking for brains... brains... brains...
 
2018-01-02 05:01:10 AM  
Poor thread. You were born doomed, and couldn't have known you would never have a chance.
 
2018-01-02 05:01:39 AM  

Mister Buttons: "Correlation
causation
scientists
reputation
demolishing
incantation
Murdochian
campaign
matter at-hand
implications
behooved
preliminary
precautions
demonstrated"


All in the first 3 paragraphs, and that's leaving out the redundant "causitives" and "correlations".


JFK
Blown away
What else do I
Have to say?
 
2018-01-02 05:03:19 AM  

randomjsa: Climate change is quite real and there's more than enough evidence now to conclude human activity is not just a major contributing factor, it is 'the' major contributing factor.

But guess what? The derpy people on the right who want to say science lies to them have a slightly less derpy cousin on the left that wants to hijack the science for personal political agenda. Why yes, I do remember the Kyoto Protocol that the left huffed and puffed over because the United States wouldn't sign up for something that would accomplish absolutely nothing in terms of climate change but would do a great job screwing the United States.

And yes, that's what comes first for the left, not the climate, but sticking it the west and the United States in particular. So here's a plan.

The left gives up on 'screw the US first, climate can come after that' plans and we come up with some good ideas and the right gives up on 'science is all a conspiracy' while we're at it.


So the right agrees it's real and a huge problem, but is pretending otherwise because ... the left haven't come up with a workable solution?  So the right have such a solution but aren't telling?  Or they don't believe they're smart enough to contribute ideas? Or ...

What on earth is your point?
 
2018-01-02 05:15:08 AM  

Flurching: LiberalConservative: starsrift: LiberalConservative: starsrift: LiberalConservative: Researches should stop (or at least reduce) conducting correlation tests and use more robust tests that better indicate causation. But correlation tests are easier and stats are hard.

What you have typed means literally nothing. Have you even ever poked your head in a high school science class?

I have taught statistics to undergrads, I have a science phd by research, and have worked has a statistician. A large proportion of statistics I see performed in science are incorrect, generally because most researchers are not statisticians. If the initial post doesn't mean much then you may be drunk (or something). Go read the basics or correlation coefficients.

If I went back to my university and told my physics prof he needed to do less "correlation tests", he'd turn into quizzicaldog.

Probably. Of the ~30 postgrads I studied with in a science lab, only 1 or 2 actually knew what they were doing with stats. Yet almost all graduated, and most continued in science (and even supervised postgrad students) though their stats knowledge was poor. Correlation is best used data exploration to identify possible relationships. Once the possible relationships of interest are identified, more robust and causation stats should then be used. If research stops at correlation then it is incomplete.

Science must predict, else it is useless.


Yes, but that is more akin to regression analysis than correlation.
 
2018-01-02 05:25:30 AM  

Flurching: wildcardjack: The theory of carbon dioxide as a greenhouse gas has been theorized and tested in lab scale fashion for over a century. The reverse applies here, causation requires correlation.

Have you heard about an IRGA (Link) it measures minute traces of CO2 using infrared absorption. This is not new science. Further, 50% of the energy we receive from the sun is in the infrared wavelengths. Just because we can't see it doesn't mean it is fake.
[img.fark.net image 425x238]


You haven't read the works of Svante Arrhenius. In the PM archives I found a reference to him to explain the weird weather of 1911, and I saw similar work by an SMU professor circa 1928.

I'm the sort to read old works full of predictions, getting laughs and shocks, such as a plasma TV display in the 1950s.
 
2018-01-02 05:28:22 AM  

ArcadianRefugee: Harlee: And that might lead to a literal conflagration. Because when science (reason) fails, there is only force and violence.

And when force is gone, there's always Mom.

/hi mom!


I am not sure if I should be proud or ashamed that I've seen Laurie Anderson in concert twice.
 
2018-01-02 05:39:59 AM  

gonegirl: ArcadianRefugee: Harlee: And that might lead to a literal conflagration. Because when science (reason) fails, there is only force and violence.

And when force is gone, there's always Mom.

/hi mom!

I am not sure if I should be proud or ashamed that I've seen Laurie Anderson in concert twice.


Proud.
 
2018-01-02 05:40:17 AM  

jso2897: Poor thread. You were born doomed, and couldn't have known you would never have a chance.


Really?  I find the thread to be quite uplifting.
 
2018-01-02 05:43:22 AM  

Dave2042: randomjsa: Climate change is quite real and there's more than enough evidence now to conclude human activity is not just a major contributing factor, it is 'the' major contributing factor.

But guess what? The derpy people on the right who want to say science lies to them have a slightly less derpy cousin on the left that wants to hijack the science for personal political agenda. Why yes, I do remember the Kyoto Protocol that the left huffed and puffed over because the United States wouldn't sign up for something that would accomplish absolutely nothing in terms of climate change but would do a great job screwing the United States.

And yes, that's what comes first for the left, not the climate, but sticking it the west and the United States in particular. So here's a plan.

The left gives up on 'screw the US first, climate can come after that' plans and we come up with some good ideas and the right gives up on 'science is all a conspiracy' while we're at it.

So the right agrees it's real and a huge problem, but is pretending otherwise because ... the left haven't come up with a workable solution?  So the right have such a solution but aren't telling?  Or they don't believe they're smart enough to contribute ideas? Or ...

What on earth is your point?


Liberals and Democrats r bad.

Don't look for more meaning than that, there isn't any.
 
2018-01-02 05:45:17 AM  

flondrix: [imgs.xkcd.com image 459x185]
Correlation doesn't imply causation, but it does waggle its eyebrows suggestively and gesture furtively while mouthing 'look over there'.


Beautifully phrased.

//also stolen
 
2018-01-02 05:49:09 AM  

LiberalConservative: Yes, but that is more akin to regression analysis than correlation


yup.
o a chi sq test and then, depending on the number of covar's, run a full non-linear second-order
 
2018-01-02 06:03:38 AM  
img.fark.netView Full Size
 
2018-01-02 06:08:16 AM  

mr lawson: LiberalConservative: Yes, but that is more akin to regression analysis than correlation

yup.
o a chi sq test and then, depending on the number of covar's, run a full non-linear second-order


Yar. All according to experimental design and test assumptions, of course.
/The zombies died, though poor loot was had.
 
2018-01-02 06:15:47 AM  

LiberalConservative: starsrift: LiberalConservative: starsrift: LiberalConservative: Researches should stop (or at least reduce) conducting correlation tests and use more robust tests that better indicate causation. But correlation tests are easier and stats are hard.

What you have typed means literally nothing. Have you even ever poked your head in a high school science class?

I have taught statistics to undergrads, I have a science phd by research, and have worked has a statistician. A large proportion of statistics I see performed in science are incorrect, generally because most researchers are not statisticians. If the initial post doesn't mean much then you may be drunk (or something). Go read the basics or correlation coefficients.

If I went back to my university and told my physics prof he needed to do less "correlation tests", he'd turn into quizzicaldog.

Probably. Of the ~30 postgrads I studied with in a science lab, only 1 or 2 actually knew what they were doing with stats. Yet almost all graduated, and most continued in science (and even supervised postgrad students) though their stats knowledge was poor. Correlation is best used data exploration to identify possible relationships. Once the possible relationships of interest are identified, more robust and causation stats should then be used. If research stops at correlation then it is incomplete.


It's not the stats themselves, its the experimental designs that determine which stats tests are appropriate. Correlational studies are cheap and many are done with existing data sets. Experimental studies (which are designed to assess causal relationships between variables) are in many disciplines - especially the social sciences and economics - either extremely expensive, unfeasible, or downright unethical.
 
2018-01-02 06:16:09 AM  

LiberalConservative: Flurching: LiberalConservative: starsrift: LiberalConservative: starsrift: LiberalConservative: Researches should stop (or at least reduce) conducting correlation tests and use more robust tests that better indicate causation. But correlation tests are easier and stats are hard.

What you have typed means literally nothing. Have you even ever poked your head in a high school science class?

I have taught statistics to undergrads, I have a science phd by research, and have worked has a statistician. A large proportion of statistics I see performed in science are incorrect, generally because most researchers are not statisticians. If the initial post doesn't mean much then you may be drunk (or something). Go read the basics or correlation coefficients.

If I went back to my university and told my physics prof he needed to do less "correlation tests", he'd turn into quizzicaldog.

Probably. Of the ~30 postgrads I studied with in a science lab, only 1 or 2 actually knew what they were doing with stats. Yet almost all graduated, and most continued in science (and even supervised postgrad students) though their stats knowledge was poor. Correlation is best used data exploration to identify possible relationships. Once the possible relationships of interest are identified, more robust and causation stats should then be used. If research stops at correlation then it is incomplete.

Science must predict, else it is useless.

Yes, but that is more akin to regression analysis than correlation.


Not necessarily, a prediction from correlated observations should be based on a model that can deliver hypotheses. The prediction may be how a system will respond IF the correlation equals causation
 
Displayed 50 of 113 comments


Oldest | « | 1 | 2 | 3 | » | Newest | Show all


View Voting Results: Smartest and Funniest

This thread is closed to new comments.

Continue Farking

On Twitter





Top Commented
Javascript is required to view headlines in widget.
  1. Links are submitted by members of the Fark community.

  2. When community members submit a link, they also write a custom headline for the story.

  3. Other Farkers comment on the links. This is the number of comments. Click here to read them.

  4. Click here to submit a link.

Report