The Games Played ‘Round the World
Rachel Bodsky
Social science experiments have traditionally been run on university campuses in North America and Europe. This results in a fairly narrow view of the world. As Joe Henrich and his colleagues Steven Heine and Ara Norenzayan point out in an article in Nature in 2010, most subjects in such experiments are "WEIRD": they come from "Western, Educated, Industrialized, Rich and Democratic" countries. It is not only that subjects tend to be university students, but perhaps more importantly that they are far from representative demographically. Henrich and his coauthors note that over 96 percent of subjects from a sample of experiments were from western industrialized countries, while only 12 percent of the world’s population is (as reported by Jeffrey Arnett, in an article in American Psychologist in 2008). This, of course, fails to be a representative sample, and calls into question the external validity of the research. This has spurred social scientists to move increasingly into the field to perform studies, resulting in new insights about how demographics affect behavior (e.g., Henrich et al, American Economic Review 2001 – see Stephanie Wang’s blog from July 30). Unfortunately, going out to the field is costly and so the scope and scale of experiments that can be performed in such a manner is limited. However, recent innovations in technology are changing that. Such studies are no longer limited to anthropologists venturing into the field. Estimates are that nearly 40 percent of the world’s population now has some sort of internet access, and the fraction of the world that is wired is rapidly growing. While access is still far from representative, it nevertheless reaches well beyond a population of western and industrialized university students. Together with Yiqing Xing, I have been running some game theory experiments using subjects recruited and paid via Amazon’s "Mechanical Turk". This is an internet-based web site where people can post various simple tasks to be performed by interested individuals from anywhere in the world for modest payments. This is a rapidly emerging way to expand and enrich a subject pool for surveys and social science experiments. Although such internet based experiments have their own selection bias, they reach well beyond the "WEIRD" subject pool and provides a wealth of interesting insights into demographics and behavior. It is clear from this experience that it is now possible to reach huge pools of subjects with a broad range of backgrounds at an affordable cost. 
There are also other sources that will enable comparisons of behaviors of people around the world. In only a few weeks since its widespread release, MobLab’s suite of games are already being played at hundreds of universities - rapidly gaining global coverage. Soon we will have tens of thousands of observations of how people with widely different backgrounds behave when put in the exactly same situation. In bringing games to the wide world, MobLab will also bring records of how people play a wide variety of games in unprecedented numbers and with incredible reach. This will not be limited to classrooms, but will soon extend to MOOCs, involving people outside of university demographics. These data will be a goldmine for the scientists conducting such research and enable them to find new insights into why people behave as they do. Are people in Beijing more or less trusting than those in Mexico City? Are people from poorer backgrounds more or less apt to cooperate in small groups than people from wealthier backgrounds? Does the religiousness of a society affect the extent to which its people contribute to public goods production? It is fun to watch MobLab grow – and it will be even more fun to see the wealth of data that it will generate. The insights that it can lead to will take cross-cultural social science research to a new level.
