/sci/ - Science & Math » Thread #13422000

114KiB, 1200x800, 1623120563478.png

View Same Google iqdb SauceNAO

Anonymous Thu 22 Jul 03:21:54 2021 No.13422000 View Reply Original Report

Quoted By: >>13422012 >>13422021 >>13422026 >>13422085 >>13422171 >>13422250 >>13422346 >>13422376

>Correlation doesn't equal causation

Anonymous

View Same Google iqdb SauceNAO 1611936188385.jpg, 71KiB, 912x1024

Anonymous Thu 22 Jul 2021 03:25:49 No.13422012 Report

Quoted By:

>>13422000
>Correlation equals causation

Anonymous

View Same Google iqdb SauceNAO EOn4ZOXW4AAETS_.jpg, 98KiB, 759x500

Anonymous Thu 22 Jul 2021 03:30:46 No.13422021 Report

Quoted By:

>>13422000
>Causation doesn't equal correlation

Anonymous

View Same Google iqdb SauceNAO Dumb2.jpg, 9KiB, 235x212

Anonymous Thu 22 Jul 2021 03:34:24 No.13422026 Report

Quoted By:

>>13422000
>equal corelation doesnt causation

Anonymous

View Same Google iqdb SauceNAO 1600926253565.png, 1010KiB, 720x720

Anonymous Thu 22 Jul 2021 03:49:18 No.13422085 Report

Quoted By:

>>13422000
Muh dick, nigga.
Oooga booga.

Anonymous

Anonymous Thu 22 Jul 2021 04:21:11 No.13422139 Report

Quoted By:

>causation doesn't correlate equality

Anonymous

Anonymous Thu 22 Jul 2021 04:32:01 No.13422171 Report

Quoted By: >>13422262

>>13422000
It's the soience equivalent of going around calling everything an ad hominem or a whataboutism when done too exuberantly, but at least it's generally not used disingenuously.

Anonymous

Anonymous Thu 22 Jul 2021 04:38:28 No.13422180 Report

Quoted By: >>13422195

The statement "correlation doesn't equal causation" is the sole reason soft sciences will never be taken serious and why soft sciences are rife with intentional and deliberate fraud.

Look at nutritional "science" for example. They often make the claim that eating red meat causes cancer. Their evidence is some study that shows a correlation between those with high in takes of red meat compared to some control.

Eating red meat may very well cause cancer. But these "scientists" are not able to explain the MECHANISTIC reason why eating red meat causes cancer.

Notice that real scientists, like physicists do not say that there is a correlation between an object being thrown and that object falling down. There is a mechanistic explanation for the theories of hard sciences.

So yes, correlation does not equal causation.

Anonymous

View Same Google iqdb SauceNAO 1623090308661.jpg, 42KiB, 680x780

Anonymous Thu 22 Jul 2021 04:41:28 No.13422184 Report

Quoted By: >>13422424

>Correlation implies causation, but by no means proves it.

Anonymous

Anonymous Thu 22 Jul 2021 04:49:58 No.13422195 Report

Quoted By:

>>13422180
Yes they need to understand the mechanism, they also need to account for the rest of hill's criteria. Proving causality in epidemiology is a complex subject.

Anonymous

View Same Google iqdb SauceNAO you.gif, 133KiB, 640x468

Anonymous Thu 22 Jul 2021 05:15:20 No.13422250 Report

Quoted By:

>>13422000

Anonymous

Anonymous Thu 22 Jul 2021 05:21:02 No.13422262 Report

Quoted By: >>13422382

>>13422171
>at least it's generally not used disingenuously.
Moron detected. Of course it's used disingenuously.
>Do you like the result?
Great! Everything is fine.
>Do you not like the result?
MUH SAMPLE SIZE
NUH CORRELATION NOT CAUSATION

Anonymous

View Same Google iqdb SauceNAO images (1).jpg, 12KiB, 300x168

Anonymous Thu 22 Jul 2021 06:03:47 No.13422346 Report

Quoted By:

>>13422000
>causation doesn't exist

Anonymous

Anonymous Thu 22 Jul 2021 06:19:44 No.13422376 Report

Quoted By: >>13422389

>>13422000
>Correlation doesn't equal causation

The most often used example is as follows: When two or more variables correlate, you may be tempted to conclude that the correlation is due to a relationship between the variables. In this instance (homicide rates go up when more ice cream is sold), there's no connection between ice cream and homicide at all: the causation is actually between temperature and ice cream, and temperature and homicide.

---

So from the above, what actually has been found is: A PROXY
Proxies are used all the time in science. Thickness of Ice deposit layers are used as a proxy for temperature. Temperature was not measured by humans a million years ago, but ice core climate proxies are a good way to get the temperature for those time periods. Tree ring proxies are used as well.

Anonymous

Anonymous Thu 22 Jul 2021 06:23:01 No.13422382 Report

Quoted By:

>>13422262
generally not used != never used
Moron.

Anonymous

Anonymous Thu 22 Jul 2021 06:28:15 No.13422389 Report

Quoted By: >>13422395

>>13422376
Continued: So from the first example, if the only data available to a scientist was ice cream sales, they could still predict homicide rates based on the data. The proxy would STILL HAVE PREDICTIVE POWER.

Proxies can be problematic though. If there were an Ice-Cream expo and and three month ice cream fare, that would skew results. But proxies are still important and should not be dismissed.

So how does correlation = causation?
Explinative power.
In other words, do you have a logical narrative to explain how this correlation results in causation?

Anonymous

Anonymous Thu 22 Jul 2021 06:35:43 No.13422395 Report

Quoted By: >>13422422

>>13422389
Continued:
If I were to tell you that there is a strong correlation between annual rainfall and average dam levels, you would say of course there is.
How did you get there? Rain falls, flows into tributaries and end up in dams.
Its obvious and logical. You accept the correlation immediately because of the logical explanation.

So when someone argues: Correlation <> causation, do you have an explanation that can account for how the two measured variables are related?

But having an explanation is not enough.
Are there competing theories?
Are they more plausible?
How strong is the original correlation?
Is it statistically significant?

Anonymous

Anonymous Thu 22 Jul 2021 06:46:21 No.13422422 Report

Quoted By: >>13422430

>>13422395
Continued:
In quantitative trading analysis, correlations are constantly generated by Quants (Quantitative analysts)
But what has the bread price in China have to do with Bicycle sales in Newark? There's a correlation (No I haven't actually checked its just an off the cuff example) It may be a proxy. It may be a proxy of a proxy ^ 60.
Most Quants will not even try to explain what's happening. They just know its a proxy and that it may hold, but when economic or other circumstances change the correlation may disappear.
So you've got a correlation, but using it to predict Bicycle sales in Newark is: USE AT YOUR OWN RISK

Anonymous

Anonymous Thu 22 Jul 2021 06:48:45 No.13422424 Report

Quoted By:

>>13422184
This.

Anonymous

Anonymous Thu 22 Jul 2021 06:53:55 No.13422430 Report

Quoted By:

>>13422422
Continued:
The conclusion is this: The difference between correlation <> causation and this correlation = causation is:
Is there a logical explanation for the correlation?
Are there competing theories?
How plausible are the alternatives?
Is there other evidence to support the correlation?
and
Could it still be valid as a good proxy?

Anonymous

Anonymous Thu 22 Jul 2021 07:34:46 No.13422512 Report

Quoted By:

>area is rife with niggers
>there is a ton of crime
>other area has few niggers
>demonstrably less crime
>ummm no sweaty correlation doesn't equal causation let me tell you about pirates in the 17th century

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches