Subtle primes of in-group and out-group affiliation change votes in a large scale field experiment

Identifying the influence of social identity over how individuals evaluate and interact with others is difficult in observational settings, prompting scholars to utilize laboratory and field experiments. These often take place in highly artificial settings or, if in the field, ask subjects to make evaluations based on little information. Here we conducted a large-scale (N = 405,179) field experiment in a real-world high-information context to test the influence of social identity. We collaborated with a popular football live score app during its poll to determine the world's best football player for the 2017-2018 season. We randomly informed users of the nationality or team affiliation of players, as opposed to just providing their names, to prime in-group status. As a result of this subtle prime, we find strong evidence of in-group favoritism based on national identity. Priming the national identity of a player increased in-group voting by 3.6% compared to receiving no information about nationality. The effect of the national identity prime is greatest among individuals reporting having a strong national identity. In contrast, we do not find evidence of in-group favoritism based on team identity. Informing individuals of players' team affiliations had no significant effect compared to not receiving any information and the effect did not vary by strength of team identity. We also find evidence of out-group derogation. Priming that a player who used to play for a user's favorite team but now plays for a rival team reduces voting for that player by between 6.1 and 6.4%.