Kind of a tangent, but I get frustrated when researchers publicly release results when one of their groups have a ridiculously small sample size. It’s irresponsible to release results based on 15 people
For example (this has nothing to do with the paper), if the average rate of some condition - some rare brain tumor - is 1 in 50M - and you take 200 COVID-19 patients and 4 of them have it - and they have several of these tumors - you can have high CI & Power with a low sample size.
When you're doing conversion rate optimization - it's usually just "did this person make a purchase?" And usually, your base rate is not less than 1/1000... But in medicine, there are ways to measure the severity of the symptoms and a lot of things are quite rare.