Causation = Correlation

Science only reaches conclusions that are supported by the statistical analysis of data. Therefore, in the end, anything we conclude regarding causation can only every be supported by correlation.

causation = correlation

prove me wrong

The problem with writing that out like an equation is that it's not reversible. The Symmetric Property of algebra doesn't apply.

is it not true if I say it the other way?

correlation = causation

Yes.

In fact, equality is not the symbol you want to use here. You want to use [math]\implies[/math]. Equality means that causation and correlation literally mean the same thing. Equality it transitive, implication is not.

how do we to determine causation then?

If A implies B, and B is not reliant on any other outside factors, then you can say B implies A.

For example,
I inject AIDS into myself. Therefore, I have AIDS.
So "Injecting AIDS" implies "You have AIDS"

However,
"You have AIDS" does not imply "I injected AIDS" because there are many other ways to get AIDS, such as through your mother or through sex. If the only way to get AIDS was injecting it into yourself, then you could say that having AIDS implied you injected yourself.

So to follow up on the scientific method, you have to try to remove or minimize outside factors that may alter the result of an experiment in order to deduce that the correlation you interpreted is indeed causing the phenomena you see. This is why many biology experiments are done in a well controlled environment such as a Petri dish.

but how do you know that getting AIDS is a result of injecting yourself with it? You can't go back in time and see whether you would still get AIDS if you decided not to inject yourself with it

Bad attempt at a counterexample.

Having AIDS is defined as having the AIDS virus in your system. Forcibly putting AIDS in your system is obviously having AIDS virus in your system.

AIDS is a syndrome that is hypothesized to be caused by a virus. The name of the virus is HIV.

How do we know HIV causes AIDS? Just because we know a lot of people with the HIV virus also happen to develop AIDS?

Also, if I didn't get AIDS from injecting myself with it, but I still had AIDS, then there must have been some external factor that gave me AIDS. But as I previously stated, in order to determine causation you have to remove all external factors or limit them (in the case of the latter, your claim of causation is more like a strongly supported conjecture).

how is it possible to ever remove all external factors?

We think HIV causes AIDS because we haven't observed any other factors that cause AIDS. If you could observe someone getting AIDS without HIV, then you would disprove the theory.

So if there are no external factors, HIV implies AIDS

We have not seen any other external factors, we assume there is none. So HIV implies AIDS unless another factor is found.

So yes, you are right, causation cannot be fully determined unless you are 100% sure that no other factors are at play. Else, you just have a strongly supported conjecture. But these things happen all the time, for example when Einstein tweaked formulations involving of the speed of light when he determined other factors at play.

You can't. This is why the concept of error exists in science.

That's the job of the scientists' experiments. It varies from experiment to experiment, examples include doing an experiment in a vacuum.

But I guess you might always be able to formulate some sort of external factor somewhere, even though the thought of it actually causing the phenomena at hand may seem outrageous. So I guess that technically no causation we've ever determined in the physical world as Humans is 100% certain, but it's pretty god damn close if we just limit those factors.

Statistical probability. Science does not make guesses. R^2 and standard deviation values are essential in writing scientific papers.

Graphs are neat but we don't simply look at a graph and say 'that kinda looks good to me'

To add on that, similar results in 2 or more standard deviations is considered statistically significant, and I think the standard for something to be considered accepted as a scientific fact is accuracy within 4+ standard deviations? It might be 5.

To put that in perspective, 3 standard deviations of a normal distribution is a 99.7% accuracy.

>R^2
>literally a measure of how well two sets of data correlate

update: it is indeed 5 deviations to be considered a fact, and this is a confidence of about 99.99994%

We never determine causation ever.

causation -> correlation (true)
correlation -> causation (false)

what is causation then? how does one determine causation?

are you just ignoring every post in the thread until you see one that agrees with your beliefs

you just quoted two different people. One of them was me, the OP.

No, I am not doing that. Are you assuming that everyone who disagrees with you is the same guy?

There is a commonly misunderstood idea that correlation does not imply causation.

This is false. When events A and B are consistently (that is to say, not just as a measurement fluke) correlated, then that does imply that there is a causal link between A and B. That is to say, it implies that there is an event C, which may or may not be equal to A or B, that causes both A and B.

The correlation does NOT imply that A causes B, or that B causes A. That's the rationale for the proverb. But the correlation DOES imply a causal link between A and B.