We experimentally test polarisation and discriminatory behaviour in Zimbabwe. We investigate how ethnolinguistic identities impact behaviours between the Shona and Ndebele people, the two major ethnolinguistic groups. Zimbabwe provides a case of deep-seated and long-term polarisation that can serve as an example for countries facing growing and increasing division. Our experiments are comprised of two main stages. The first stage of the experiments involved participants undertaking a variety of within-subject experimental games using well-documented experiments in the literature: the prisoner's dilemma, dictator game, trust game and public goods game. In each game, we reveal the receiver's identity as Ndebele, Shona and Zimbabwean citizen. In other words, participants have to distribute their endowment to an in-group, out-group and 'mixed-group' member. Participants were informed that the Zimbabwean receiver has a 50% probability that the individual is Shona or Ndebele.

Following the games, individuals participate in a modified version of the social norms elicitation task. We also elicited participants' beliefs regarding others' decisions for all social dilemma games. In our social norms elicitation task, we asked participants to evaluate the social appropriateness of various behaviours in different real-world scenarios. Lastly, participants complete a post-experiment questionnaire to identify demographic characteristics and self-reported identity salience to measure the extent of group affiliation further. In the second stage, participants are randomly assigned to one of the two norm-based treatments and a control group who receive no information to evaluate how behaviour can be shifted. Using a proscriptive social norm, we test if discriminatory behaviour can be further exacerbated. In an alternative treatment, we use a prescriptive social norm nudge to test if discriminatory behaviour can be minimised.

We find evidence of discriminatory behaviour. The findings suggest that the frequency of interaction and the strength of identification play a role in discriminatory behaviour. In addition, beliefs largely determine behaviours. We observe treatment effects following the norm-nudge treatments. Both the proscriptive and prescriptive treatments yield robust results, but the direction of the impact varies across different choice contexts. The regression analysis reveals that Ndebele participants generally respond positively to both treatments, increasing cooperation for in-group, out-group and Zimbabwean receivers. However, the prescriptive treatment backfires by reducing cooperation from Shona players to Ndebele receivers. Our findings highlight that minority groups may be more willing to cooperate and interact with the majority group, whilst the majority group may show greater signs of hostility.