To understand discrimination in cooperation researchers use experiments — among them dictator and trust games. In such experiments players do not know more about other participants than what the researcher tells. For example, a researcher interested in migrant-native cooperation would design an experiment in which the players only learn about the migration status of the other player. An important feature of social interaction, however, is that individuals often make more “informed” decisions. In the developing world rural villagers know each other well. First, they know more about each other than only the attribute under study. Second, their behavior is guided by previous interaction. Finally, their interaction depends on their network: their relationship with other, third actors.

We implement a set of novel dictator and trust games to make four contributions.

First, the identity of the other player is revealed to the participant via an instant photo. This allows players to make fully informed decisions. It also allows the researcher to analyze receiver characteristics beyond the attribute under study. Second, each participant plays the dictator game in two roles: as sender, and in the role of receiver as many times as there other players. This so-called “round-robin” structure allows us to directly observe two-way interaction. Third, by playing a set of dictator and trust games among individuals that know each other well we obtain a behavioral measure of sharing and trust networks. Finally, we stratify our sample upon migration status, and analyze native-migrant cooperation. Due to conflict and high levels of displacement this is a key cleavage in the Congolese context.


Download the complete dataset for all 416 players. Download.


  • IRB: IRB-AAAJ2401. December 17, 2011.
  • Local Approval.  PDF.
  • Player Survey. PDF.
  • Team Leader Survey. PDF.
  • Protocol and Codesheet. PDF.


  • Team: N. Sircar and P. van der Windt.
  • Partners: None.
  • Funding: ASC and CSDS.
  • Location: South Kivu, DR Congo.
  • Participants: 416 players, 24 villages.


We needed three isolated areas to play the experiments. Normally we confiscated a school or a church in a village, and used plastic sheeting to create isolated areas. In one village this was not possible so we confiscated three shops. One enumerator played the games in the public phone booth, another played in the pharmacy and the third enumerator played in yet another shop (not on the picture).


The enumerators revealed the identity of the other participants via an instant photograph.


Per village we needed to print 54 photos. Because the printers consumed a lot of battery, upon entry in a village one of our first questions was “Is there is a generator in the village?”. If not one enumerator would spend the rest of the day finding one in other villages. We then bought two liters of petrol and half a liter of oil (around $5 all together) which was enough for 5 hours of electricity.


In one village, after a long search, we found one generator that had clearly not been used in months (if not years). Once we put the oil and petrol in and connected the printers, the family that owned the generators took the protection cover of the television. When we came back two hours later, almost the whole village was present watching television. So, if the research has no impact whatsoever, it had at least one positive external effect.