@tomstafford one of the first thing every student/researcher is taught is the importance of random sampling and randomization... and then they go ignore all that and just pretend it's all good (for instance, you can't directly compute p-values in observational studies as the null distribution is not known)