Addendum to "Rule o' Thumb"

I earlier wrote about the statement that correlation does not imply causality:

That's correct. Given that in the social sciences (with the exception of psychology, assuming you call that a social science) you can rarely run an experiment, it's also not very helpful. The rule of thumb that I use is that correlation implies causation to the extent that alternative explanations for the correlation are implausible.
I should have added two points to that post:

1. The very basic problem is that one cannot observe causality at all - you cannot see, hear, smell, feel or taste causality. When you clap your hands a thousand times and hear a clapping sound every time and never hear that sound when you don't clap your hands, that's a correlation. Causality is a concept that your brain uses to interpret that correlation. I believe Kant said something similar.

2. Against this backdrop, randomized experiments, the gold standard in the social sciences, can be understood as an attempt to minimize the number of interpretations of correlation other than causation. If you run a randomized experiment and you get a different result for the treatment than the control condition, there are only three explanations:

a) You somehow messed up.

b) Chance

c) Causality

That's as good as it gets.

No comments: