How to Randomly Sample the Globe: A First Step

There is a major gap between collecting convenience samples in western countries and a simple random sample of humans all over the world which would be necessary for the generalization of research findings. But, although research might never reach the generalizability of the globe, it does not mean we can not move closer. In this article, we map a framework for moving closer towards true generalizability of human subjects research. It starts with drawing cluster random samples using countries as clusters, drawn proportional to their populations. Then, through Big team Science, reaching out and building connections at those sites for participation. Finally, correcting treatment effects for nonresponse bias at the cluster (country) level through country-level indicators. This framework starts to bring us closer to random sampling of the globe while correcting for nonresponse bias at the cluster level. Future work can refine this procedure and make further steps towards generalizable human sciences.