Hacker News Reader

If all laymen knew that n is irrelevant, it's the effect size that matters, we would be in a world of good.

If the difference between means is large compared to the standard deviation of the population under study, you need less n.

Another fun fact: All groups are different given large enough n.

Edit: I'm not endorsing the study - just pointing out that n should not be the most critical thing you look for when evaluating a study. The study is fine as a simple quasi-experiment, testing the hypothesis by induction. I don't see anyone claiming causal links here. Now a study testing the hypothesis deductively needs to be designed.

Thank you for body slamming me. But in a nice way. I stand corrected.

I think the school, that conducted the study had some effect on it being published.

I don't suspect any school below a rank of 15 would be taken seriously.

+1 for your username