4 ms·
Wow, is it really normal to publish results from such small sample sizes? This feels like tossing dice and publishing the result.
by mlang23 5y ago
Wow, is it really normal to publish results from such small sample sizes? This feels like tossing dice and publishing the result.
- mountainb 5y agoYeah. There are other tricks to make a tiny n seem more relevant, like making the relevant (tiny) n a subset of a bigger sample, taking many separate samples from different populations and looking for correlations in them separately, and so on. Trust the science.* *Whereas science is defined as getting a certain score in an abstract statistics game.
- mlang23 5y agoWhat I find most problematic is that there is clearly a mix of difference incentives at play. At the one hand, scientists are not getting tired of waving the scientific method about to the uninitiated. Claiming science has no agenda and is only interested in hard data. But on the other hand, it is rather clear that the career of individual scientists depends on how much they publish. Yes, there is peer review, but no, that doesnt help, because people form networks and have friends and allies. So, how do we ensure that the common people trust science more? I think we should start by ridiculing publications like this, if the scientific community by itself is not capable of filtering such noise out.
- sideshowb 5y agoYes. Though it's also normal that whatever your sample size, someone will always complain it wasn't bigger.