Optimistically, if people from across some kind of ethnic or cultural divide spend more time interacting with each other on a human level they’ll learn to get along better. Pessimistically, familiar breeds contempt. Via Tyler Cowen, Sharon Barnhardt offers some evidence for the optimistic view:
This paper provides experimental evidence on whether religiously diverse neighbors affect attitudes about another religious group and/or preferences for inter-religious living. I exploit a natural experiment in a large Indian city in which Hindus and Muslims were randomly assigned units in a public housing complex with physically distinct “clusters” of four units. The lottery generates exogenous variation in the degree of ethnic diversity across clusters within the complex, which is rare among adult households. I conduct an original survey of 1363 households focusing on attitudes about members of the other religion and willingness to live together. To overcome concerns about self-presentation bias, I also measure implicit attitudes via an Implicit Association Test (IAT) for a representative sub-sample. My estimates demonstrate location influences interactions in that individuals spend significant time with others in their cluster. Increased proximity and interaction, in turn, affect attitudes. Greater exposure to Muslims (the minority group) improves Hindus’ explicit attitudes about Muslims by 0.25 to 0.40 standard deviations, depending on the measure, and increases their willingness to live with Muslims. Paralleling this, I observe significant reductions in implicit bias (0.20 to 0.57 standard deviations) among Hindu children. While I observe no significant effects for Muslims, the overall effect is a convergence of attitudes across religious groups. As India expands public housing for the poor to accommodate rapid urbanization, deliberate mixing of religious groups can be a way of improving attitudes toward the religious minority.
Ideally you’d want to have someone go around the world looking for other examples of basically similar “natural experiments” to see how well this holds up. Unfortunately, I think the incentive structure is for graduate students to do something more original than “let’s do the same thing as what Sharon Barnhardt did that one time and see if we get similar results.”