Remember, correlation is not causation. Causation can be determined by experiments (either "natural experiments" or intervention studies), but obviously these

Remember, correlation is not causation. Causation can be determined by experiments (either "natural experiments" or intervention studies), but obviously these are both technically impossible and usually immoral to do on humans. Some fields like (…parts of) economics and medicine do experiment design properly, but many others (genetics, psychology, nutrition) just kind of ignore it and hand-wave something about P-values. This is ok if you're only using genetics to create 23AndMe ancestry charts, but it won't be once we actually have gene editing capabilities to inform.