They are science, but they're not good evidence. A well run meta analysis of well run RCTs would be best, followed by well run RCTs. Anything else is science b

They are science, but they're not good evidence. A well run meta analysis of well run RCTs would be best, followed by well run RCTs. Anything else is science but weak evidence.