Can someone Eli5 please.
In: 8
Correlation is a trend where something occurs and you see that hey. This other things tends to occur more or less. But it doesn’t mean that one actually causes this to occur.
My fav example. Correlation between margarine consumption and divorce. The more margarine consumed. The greater the divorce rate. https://www.tylervigen.com/spurious-correlations
Causation on the other hand. If I take 100 people. I believe that putting your hand on a hot stove will cause a burn. I put it to the test. Yes. 100 people burned their hands. And it was caused by the heat from the stove.
That being said, correlations are useful and necessary. We’ll defined variables/covariates and modeling. We can’t always conduct the cause and effect studies. But we can reasonably infer that there is a strong relationship between two (or more) variables.
Causation represents a mechanism that drives something to happen. You can find causes in things like natural laws governing interactions of matter and energy, human psychology, etc. These things are INHERENT to the subjects that is doing the action, regardless of the existence and manner of an observer
Correlation is a statistical phenomena that us human beings observe, using a particular kind of abstraction we chose to measure and represent the things that interest us. We create ways of numerically measuring things, the units that represent these measurements, and we plot a series of measurements on a Cartesian coordinate system and we go “look at that, these two lines sure look close to each other”. The concept of “correlation” does not exist, without us observing and describing in a particular way.
Thank you all, it means what I thought but you all helped break it down in terms I can grasp.
Causation represents a mechanism that drives something to happen. You can find causes in things like natural laws governing interactions of matter and energy, human psychology, etc. These things are INHERENT to the subjects that is doing the action, regardless of the existence and manner of an observer
Correlation is a statistical phenomena that us human beings observe, using a particular kind of abstraction we chose to measure and represent the things that interest us. We create ways of numerically measuring things, the units that represent these measurements, and we plot a series of measurements on a Cartesian coordinate system and we go “look at that, these two lines sure look close to each other”. The concept of “correlation” does not exist, without us observing and describing in a particular way.
The two are not exclusive.
“Correlation between A and B” means A and B both tend to happen at the same time (or to the same people, etc). It may be because A directly or indirectly causes B. Or it may be because B causes A. Or it may be coincidence, or it may be because something else (call it C) causes A and B.
So when people say “Correlation doesn’t mean causation”, what they’re saying is, if you notice A and B are correlated, don’t automatically assume it’s because A causes B.
For example, you may observe that smokers tend to get lung cancer. That’s correlation. Is it because smoking increases the risk of lung cancer? In this case, yes. It’s causation.
You may also observe that people who carry cigarette lighters tend to get lung cancer. Thats’ also a perfectly valid correlation. Is it because cigarette lighters cause lung cancer? Probably not. It’s because something else (smoking) causes lung cancer and also causes you to carry a cigarette lighter. So in this case, it’s correlation but not causation.
Sometimes it’s very hard to distinguish between the two, which is why it’s so important to find out. For example, if a study shows people who drink red wine live longer than people who drink beer, you’d be tempted to concluded that red wine is good for you. But it might be because the kind of people who drink red wine tend to be middle-class people who can afford better health care, while beer is more popular among blue-collar workers.
The two are not exclusive.
“Correlation between A and B” means A and B both tend to happen at the same time (or to the same people, etc). It may be because A directly or indirectly causes B. Or it may be because B causes A. Or it may be coincidence, or it may be because something else (call it C) causes A and B.
So when people say “Correlation doesn’t mean causation”, what they’re saying is, if you notice A and B are correlated, don’t automatically assume it’s because A causes B.
For example, you may observe that smokers tend to get lung cancer. That’s correlation. Is it because smoking increases the risk of lung cancer? In this case, yes. It’s causation.
You may also observe that people who carry cigarette lighters tend to get lung cancer. Thats’ also a perfectly valid correlation. Is it because cigarette lighters cause lung cancer? Probably not. It’s because something else (smoking) causes lung cancer and also causes you to carry a cigarette lighter. So in this case, it’s correlation but not causation.
Sometimes it’s very hard to distinguish between the two, which is why it’s so important to find out. For example, if a study shows people who drink red wine live longer than people who drink beer, you’d be tempted to concluded that red wine is good for you. But it might be because the kind of people who drink red wine tend to be middle-class people who can afford better health care, while beer is more popular among blue-collar workers.
A caused B is causation. An object thrown straight up falls back down due to gravity is causation, one action directly follows the other.
C happens at the same time as A caused B but being unrelated is causation. You throw an object up and it falls down, and at the same time the object is truck by lightning. You didn’t cause the lightning, it just happened at the same time and place as something you did cause.
One of the most common logical fallacies is ‘Post Hoc, ergo propter hoc’ which translates to ‘After it, therefore because of it’. This statement often confuses causation with correlation, as the assumption is any event B that occurs after event A was caused by it, instead of the possibility that Event B was only correlated, not caused, by Event A.
A caused B is causation. An object thrown straight up falls back down due to gravity is causation, one action directly follows the other.
C happens at the same time as A caused B but being unrelated is causation. You throw an object up and it falls down, and at the same time the object is truck by lightning. You didn’t cause the lightning, it just happened at the same time and place as something you did cause.
One of the most common logical fallacies is ‘Post Hoc, ergo propter hoc’ which translates to ‘After it, therefore because of it’. This statement often confuses causation with correlation, as the assumption is any event B that occurs after event A was caused by it, instead of the possibility that Event B was only correlated, not caused, by Event A.
Just to add to other good explanation, I once heard this one and fee like it explains it very well:
During the summer, people buy bore icecream. People also get bitten by a shark more often in the same period. You could then say icecreams cause sharks to bite people. They don’t.
Just to add to other good explanation, I once heard this one and fee like it explains it very well:
During the summer, people buy bore icecream. People also get bitten by a shark more often in the same period. You could then say icecreams cause sharks to bite people. They don’t.
Latest Answers