I don’t believe you can obtain an understanding of the idea that “correlation does not imply causation” from even a very deep appreciation of the material in Statistics 101. These courses usually make no attempt to define confounding, comparability etc. If they try to define confounding, they tend to use incoherent criteria based on changes in the estimate. Any understanding is almost certainly going to have to originate from outside of Statistics 101; unless you take a course on causal inference based on directed acyclic graphs it will be very challenging to get beyond memorizing the teacher’s password
Agree completely, and I’ll also point out that at least for me, a very shallow understanding of the ideas in Causality did much more to help me understand correlation vs. causation, confounding etc. than any amount of work with Statistics 101. And this was enormously practical–I was able to make significantly better financial decisions at Fundation due to understanding concepts like Simpson’s Paradox on a system 1 level.
To chime in as well: my own understanding of ‘correlation does not imply causation’ does not come from the basic statistics courses and articles and tutorials I read. While I knew the saying and the concepts and a little bit about causal graphs, it took years of failed self-experiments and the intensely frustrating experience of seeing correlate after correlate fail randomized experiments before I truly accepted it.
I don’t know how helpful, exactly, this has been on a practical level, but at least it’s good for me on an epistemic level in that I have since accepted many fewer new beliefs than I would otherwise have.
Although you know, there is no reason in principle you couldn’t get all that stuff Anders_H is talking about from intro stats, it’s just that stats isn’t taught as well as it can be.
I don’t believe you can obtain an understanding of the idea that “correlation does not imply causation” from even a very deep appreciation of the material in Statistics 101. These courses usually make no attempt to define confounding, comparability etc. If they try to define confounding, they tend to use incoherent criteria based on changes in the estimate. Any understanding is almost certainly going to have to originate from outside of Statistics 101; unless you take a course on causal inference based on directed acyclic graphs it will be very challenging to get beyond memorizing the teacher’s password
Agree completely, and I’ll also point out that at least for me, a very shallow understanding of the ideas in Causality did much more to help me understand correlation vs. causation, confounding etc. than any amount of work with Statistics 101. And this was enormously practical–I was able to make significantly better financial decisions at Fundation due to understanding concepts like Simpson’s Paradox on a system 1 level.
To chime in as well: my own understanding of ‘correlation does not imply causation’ does not come from the basic statistics courses and articles and tutorials I read. While I knew the saying and the concepts and a little bit about causal graphs, it took years of failed self-experiments and the intensely frustrating experience of seeing correlate after correlate fail randomized experiments before I truly accepted it.
I don’t know how helpful, exactly, this has been on a practical level, but at least it’s good for me on an epistemic level in that I have since accepted many fewer new beliefs than I would otherwise have.
Me four.
Although you know, there is no reason in principle you couldn’t get all that stuff Anders_H is talking about from intro stats, it’s just that stats isn’t taught as well as it can be.