Social trust is considered key for social, political, and economic development. But what does social trust look like in the real world? Who trusts whom, and with what? We seek to answer these questions using data from n=21,000 randomized interactions in 7 countries across the world: Argentina, Bangladesh, India, Kenya, Lebanon, Pakistan, and the U.S. These interactions measure patterns of cooperation among strangers from different social groups. With a team of confederates, we randomly select pedestrians, and manipulate: (1) the type of interaction they are faced with — and thus the level of trust required (dropping groceries, asking for directions, asking to borrow a cell phone); (2) the confederates’ gender; (3) the confederates’ ethnicity; and (4) the confederates’ socio-economic status. In every country, in every experiment, gender is the strongest determinant of both trust and trustworthiness --- far outweighing the roles of class or ethnicity. We find that women are much less likely to trust strangers --- even other women --- relative to men. This result runs counter to a robust finding from lab games and surveys, which consistently show that women are more trusting than men. We propose that safety concerns and regressive social norms drive this result. We discuss the implications of the gender gap in social trust for building broader social cohesion.

Written with Saad Gulzar, Princeton University