And on the extreme end you have the multi-decade longitudinal studies in epidemiology / biomedicine that would be more-or-less impossible to replicate.
I remember reading that some epidemiologists saw the wealth of new data from CoVID as a silver lining because of how few events there are at that scale. Apparently it’s not uncommon to still use the Spanish Flu data which is spotty at best because it might be the only thing available at the scale you’re interested in
