ABSTRACT
Previous research has claimed that world religions can extend the in-group beyond local and ethnic boundaries to form larger multi-ethnic groups, expanding human societies. Two experiments were run in Fiji to test religion’s ability to expand group boundaries. Experiment 1 employed a religious prime to increase prosocial behavior towards co-religionists among Hindu Indo-Fijians in an economic game. There were no overall effects of priming, but gender-specific effects were found. Priming reduced the amount women biased coin allocations to favor their preferred group. Men showed no bias in either condition. Experiment 2 employed the same economic game, without a prime, in a sample of indigenous Fijian and Indo-Fijian Christians. In this game, the monetary allocations were made between different religious and ethnic groups to test if preferences for religious in-groups were stronger than preferences for ethnic in-groups. Indo-Fijian Christians showed bias against their own ethnic group if they were from a different religion (Hindus or Muslims), but allocated fairly towards Christians from a different ethnic group (indigenous Fijians). Indigenous Fijians allocated less money to Muslims, but not Hindus. This evidence suggests that religious bonds can overcome the preference for one’s own ethnic group and expand in-groups to multi-ethnic religious groups.
Acknowledgments
I would like to thank Benjamin Purzycki, Dimitris Xygalatas, Coren Apicella, Joseph Henrich, Ara Norenzayan, Justin Busch, Adam Baimel, and Nicole Wen for their very helpful comments on this article.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1. All coins were distributed to members of the appropriate group, local and distant, after the games were completed.
2. Though attempts were made to keep the money counting and payment anonymous by presenting money only in sealed numbered envelopes, almost every participant insisted on thanking the money counter personally after the money was given and it proved difficult to stop them.
3. Collusion in the form of organizing to “beat” the game did not appear to be a problem, but gossip definitely presented some issues. Neither I, my research assistants, nor any of our informants overheard anyone in the villages talking about the game itself outside of the game setting, but the villagers were very interested in our presence in the village and the money we were giving to participants. I correlated co-religionist allocation with participant order to check for collusion. If collusion was present, cup allocation should change over time. Correlations between subject number and allocation were not significant, suggesting this was not the case (Self Game: r = −0.02, 95% CI −0.24 to 0.21; Local Co-Religionist Game 2: r = 0.13, 95% CI −0.10 to 0.35). In casual interviews with participants after the game, it became clear that the main theory as to our presence was that we were from the Indian government to help the poor Hindus. Our assurances that this was not the case did little to discourage this belief. Though it is unclear if or how this affected the game, it may be the case that people behaved more fairly than they otherwise would, based on the belief that they were to act as “good Hindus” to garner favor for foreign government aid.
4. Adding random slopes into these equations does not meaningfully change the effects or effect sizes, but makes both of the effects for the Muslim games marginally significant in .
5. It is worth noting that some of this difference may be due to male participants allocating more to themselves than female participants in the Self Game. Even though this difference is non-significant, it can still have some impact on the interaction effects. The same logic applies to the interaction effect in the indigenous Fijian games, where men allocate less to themselves than women in the Self Game.