# Correlation and Causation

Discussion in 'Mathematics' started by stevencarrwork, Jan 24, 2011.

1. ### stevencarrwork

Knowing that correlation is not the same as causation correlates very highly with taking a course in statistics.

What does that prove exactly?

2. ### stevencarrwork

Knowing that correlation is not the same as causation correlates very highly with taking a course in statistics.

What does that prove exactly?

True
How come?
Dunno

4. ### NazardNew commenter

Prove? Nothing at all (given that this is stats and I don't think proof comes into stats).
If there is correlation between the two parts of the first sentence then this doesn't necessarily mean that taking a course in stats give one the requisite knowledge (because this would be assuming causality).
So: I am going to suggest that "Knowing that correlation is not the same as causation correlates very highly with taking a course in statistics" doesn't disprove the suggestion that learning a course has no bearing on how well you understand the material in that course.
I think.
How did I do?
[I thought this was obvious when I read it first - now I'm not at all sure!]

5. ### GoldMathsNew commenter

What came first the chicken or the egg?

6. ### googolplexOccasional commenter

According to QI, it was the egg, cos chickens would have evolved from reptiles which also came from eggs; so eggs came before chickens. The explanation left me thinking "what came first, the reptile or the egg...?"

7. ### ResourceFinder

It is always the egg

Since something that is not a "blah" lays the egg that hatches the "blah"

That's evolution

8. ### DMNew commenter

Where's our resident evolutionary bioinformatician when you need him?
bgy1???

9. ### bgy1mm

Here.
We can easily posit a causal mechanism - which is that statistics courses teach students that correlation doesn't necessarily imply causation.
However there are possible alternate hypotheses. One is that intelligent students, who know that correlation doesn't imply causation, gravitate towards statistics courses. Whilst we could test this - by taking a sample of students before they take their course - we'd need rather a large cohort because we can't predict in advance who will take the course and who won't. You can't necessarily research everything.
There are other things we can do, such as test that correlation doesn't equal causation is actually taught, and test that other subjects that are taught - eg how to calculate a chi-square, are actually learnt.This tells us whether teaching as a whole is effective.
Ultimately you can only reject hypotheses in statistics, you can't prove them.