>And I'm pretty confident the majority of experiments are using sample sizes larger than 12. It definitely varies by field, testing 3 doses and a negative cont

>And I'm pretty confident the majority of experiments are using sample sizes larger than 12. It definitely varies by field, testing 3 doses and a negative control with 3x replication would be a pretty good experiment for testing a drug in mice. It's plenty of data for identifying a drug that actually works, as long as your statistical tools aren't designed with expectation of thousands of samples.