Experimental Design Details
The Contact treatment resembles the protocol developed by Aron et al. (1997): subjects are paired with an assistant, and take turns in answering a series of questions. The questions, directly drawn from the ``Closeness-Generating Procedure'' in Aron et al. (1997), gradually become more self-disclosing and relationship-building. The first questions are fairly neutral, to get the procedure started (e.g. ``Before making a telephone call, do you ever rehearse what you are going to say? Why?''). The second set of questions then aims to reveal more-personal details (e.g. ``What do you value most in a friendship?'', ``How do you feel about your relationship with your mother?''). Last, the third set covers very personal topics, to create a strong link between the two individuals (e.g. ``If you were going to become a close friend with your partner, please share what would be important for him or her to know''. The subject and the assistant were both asked to answer these questions (without any supervision from experimenters).
In the Photo treatment, subjects are again paired with one of the assistants from the Contact treatment. However, they do not meet them, but are only shown their photo with their names - which include considerable information about ethnicity Mabudike (1976).
In both the Contact and the Photo treatments, subjects were explicitly told that the games played during the experiment will be with the person presented (either in person or in the photo).
In the Anonymous treatment, subjects are not presented the person with whom they will play, and are rather told that they are going to play with ``someone selected at random''.
We decided to select assistants from outside the experimental subject pool - as opposed, for instance, to other subjects - for several reasons. First, because they will meet multiple subjects, we can introduce individual assistant fixed effects in the Contact and Photo treatments. Second, this was one way of ensuring that at least one person in the pair was literate, and therefore did not require a third person to supervise the discussions (which could have influenced the results). Third, we used assistants for feasibility reasons, as we used a paper format for the photos in the Photo treatment and so required photos in advance. Last, having assistants as pairs allowed us to randomize meetings with respect to ethnicity, which would have been much more difficult otherwise, especially in the fishing village which is almost entirely ethnically homogeneous.