What's that quote about correlation and causation?

Current Events

Page of 2
Current Events » What's that quote about correlation and causation?
The_cranky_hermit posted...
"Coincidence does not imply causation."

1-3 data points is not a correlation. Coincidence is not a true correlation. You may mistake it for a correlation, but when it lacks predictive power you'll be able to confirm there is no correlation.

pinky0926 posted...
Correlation does not mean causation, but often it implies it

Not really. Correlation means there is a relationship. Causation is one of the many ways it can be related.

For example, about 10 years ago I looked at every US city with a population over 300,000 and compared the median income to homicide rate. There were low crime poor cities, high crime poor cities, and low crime rich cities. There were no high crime rich cities. With a sample that large, if the two variables were unrelated then there should be data points in all 4 corners. I'm confident that data showed a relationship between crime and income. There was a correlation.

But it's not causation. I cannot say that poverty leads to more people choosing a life a crime, that would be causation. I cannot say that more crime harms businesses and causes poverty, that would be causation in the other direction. Nor can I say they will always be related as other situations change. But in 2015, there was a correlation, those two factors were somehow related, and I suspect that correlation still exists today.
"Something's wrong! Murder isn't working and that's all we're good at." ~Futurama
The_cranky_hermit posted...
You said you'd disprove any link.
further research then disproves any causal link
...you're taking the piss, right? You're conflating "any" with "every", which I can only assume is intentional because you are irritatingly specific with your wording.

To use another silly example, there's a strong correlation between the distance between Neptune and the Sun, the viewership for "Days of our Lives" and the remaining Forest cover in the Brazilian Amazon. Now, we know that there is no causal link between any of these because there is literally no way for these to affect each other. Any .

I believe this has been a waste of my time because you are not here in good faith.
Fan of metal? Don't mind covers? Check out my youtube and give me some feedback
http://www.youtube.com/sircaballero
reincarnator07 posted...
To use another silly example, there's a strong correlation between the distance between Neptune and the Sun, the viewership for "Days of our Lives" and the remaining Forest cover in the Brazilian Amazon. Now, we know that there is no causal link between any of these because there is literally no way for these to affect each other. Any.

Correlation value r tells you how well the given data points follow a trend and it can give a false impression of a strong relationship based on a coincidence. But going deeper into statistics, there are ways to deal with that. Next you'd want to look at the significance from the p-value which takes into account the number of data points. The p-value looks at the likelihood of a random number generator getting these results which is a huge problem if you have only a couple data points. Finally you'd want to look at predictive power as a way to combat cherry picking. Expand out your Days of Our Lives vs Sun to Neptune Distance vs Forest Acreage data set to include more years and I highly doubt it would remain a strong correlation.

TLDR: Those don't actually have a strong correlation.
"Something's wrong! Murder isn't working and that's all we're good at." ~Futurama
bfslick50 posted...
1-3 data points is not a correlation. Coincidence is not a true correlation. You may mistake it for a correlation, but when it lacks predictive power you'll be able to confirm there is no correlation.

Not really. Correlation means there is a relationship. Causation is one of the many ways it can be related.

For example, about 10 years ago I looked at every US city with a population over 300,000 and compared the median income to homicide rate. There were low crime poor cities, high crime poor cities, and low crime rich cities. There were no high crime rich cities. With a sample that large, if the two variables were unrelated then there should be data points in all 4 corners. I'm confident that data showed a relationship between crime and income. There was a correlation.

But it's not causation. I cannot say that poverty leads to more people choosing a life a crime, that would be causation. I cannot say that more crime harms businesses and causes poverty, that would be causation in the other direction. Nor can I say they will always be related as other situations change. But in 2015, there was a correlation, those two factors were somehow related, and I suspect that correlation still exists today.

Ah ok so it...implies it
CE's Resident Scotsman.
http://i.imgur.com/ILz2ZbV.jpg
pinky0926 posted...
Ah ok so it...implies it
So, in discourse, implication means that you're making a suggestion, but that's not what's meant here. Logically speaking, "implication" means it's impossible for the cause to be true and the effect to be false. So no, implication doesn't imply causation. It might hint at it, but it's possible for the correlation to be there but the cause to actually be something else.
Arguing on CE be all like:
https://youtu.be/JpRKrs67lOs?si=kPGA2RCKVHTdbVrJ
reincarnator07 posted...
You're conflating "any" with "every", which I can only assume is intentional because you are irritatingly specific with your wording.
Yes, and? "I can refute any link" is equivalent to "I can refute every link." It is not equivalent to "I can refute one link." And I was being specific with my wording on every single post I made regarding your claim, precisely so that you couldn't weasel out of it by changing its meaning (as you just tried to).

ReturnOfDevsman posted...
Logically speaking, "implication" means it's impossible for the cause to be true and the effect to be false. So no, implication doesn't imply causation.
Science isn't a logic puzzle. Neither is statistics. These deal in probabilities, not absolute truth values. A statement like "correlation implies causation" can have exceptions and still be generally correct.

Also, logically speaking, there's a difference between "correlation implies causation" and "correlation always implies causation." And a difference between "correlation doesn't imply causation" and "correlation never implies causation."

it's possible for the correlation to be there but the cause to actually be something else.
That would still count as implying a cause. The correlation isn't the cause, but it implies another one.
http://thecrankyhermit.shoutwiki.com
Year-by-year analysis of the finest gaming has to offer, and (eventually) more!
The_cranky_hermit posted...
Science isn't a logic puzzle. Neither is statistics. These deal in probabilities, not absolute truth values. A statement like "correlation implies causation" can have exceptions and still be generally correct.
I should have known better than to expect CE to be interested in anything other than winning an Internet fight.
Arguing on CE be all like:
https://youtu.be/JpRKrs67lOs?si=kPGA2RCKVHTdbVrJ
Post #58 was unavailable or deleted.
The_cranky_hermit posted...
Yes, and? "I can refute any link" is equivalent to "I can refute every link." It is not equivalent to "I can refute one link."
That is rich when in the same post you wrote

Also, logically speaking, there's a difference between "correlation implies causation" and "correlation always implies causation." And a difference between "correlation doesn't imply causation" and "correlation never implies causation."
If you want to strawman me to win an internet argument, then save the electricity and do it in your head
Fan of metal? Don't mind covers? Check out my youtube and give me some feedback
http://www.youtube.com/sircaballero
Remember when people were burning down 5G cell towers to prevent COVID? Many of the conspiracy theories arise due to correlation.
There's also the classic: "As the number of pirates decreased, the average global temperature has increased".
Home of the free
Because of the brave
reincarnator07 posted...
That is rich when in the same post you wrote [a side-note about semiformal logic]
You missed the part where I said logical definitions aren't applicable here. I'm not issuing a directive to differentiate between those statements. I'm just saying that under a logical system, they would be.

JTsyo posted...
Remember when people were burning down 5G cell towers to prevent COVID? Many of the conspiracy theories arise due to correlation.
5G and COVID were never actually correlated in any meaningful sense of the word.
http://thecrankyhermit.shoutwiki.com
Year-by-year analysis of the finest gaming has to offer, and (eventually) more!
JTsyo posted...
There's also the classic: "As the number of pirates decreased, the average global temperature has increased".
There's not a particularly strong correlation there, unless you pick and choose. You have far more samples where there are zero pirates and the global temperature is lower than today, than you have samples where there is any non-zero number of pirates, or higher temperatures.
Arguing on CE be all like:
https://youtu.be/JpRKrs67lOs?si=kPGA2RCKVHTdbVrJ
Current Events » What's that quote about correlation and causation?
Page of 2