Introduction. For the first time, Xie et al. (Nature Human Behaviour, 1(8), 0142. DOI: 10.1038/s41562-017-0142, 2017) found that the probability that disinterested third parties choose to reduce inequality was lower when it also involved inverting recipients’ relative payoff positions, a phenomenon which they called Rank Reversal Aversion (hereafter RRA). This phenomenon posits questions in terms of how the preference for a stable hierarchy might be traded off against conflicting values, such as equality and fairness.

Goal. The goal of the present experiment was to assess participants´ distributive preferences when RRA, Inequality Aversion, and fairness considerations were in conflict.

Methods. We present a pre-registered experiment (https://osf.io/8kdx6) with university students in which we propose a methodologically cleaner alternative to explore whether people may have RRA. Some participants played a Third-Party Dictator Game with real consequences in which they had to decide on monetary allocations destined to two participants in roles A and B. The experiment had a within-subject design in which each dictator faced 40 allocation choices. To asses RRA, some choice trials involved an unequal initial endowment for A and B, while the Dictator could make monetary transfers that changed A and B´s initial payoff rank without changing the inequality between them. To assess the importance participants attributed to inequality over rank, there were choices in which preserving the initial rank led to slight increases of inequality. In turn, to test for fairness, in some trials the dictator was also provided with information about A and B´s relative performance in an effortful task.

Results. When dictators´ transfers could not alter the inequality between A and B and fairness was not at stake (there was no information about performance), results confirmed a RRA: participants were reluctant to change A and B´s initial payoff rank in 65% of these choices. However, when RRA was pitted against eliminating inequality, participants preferred to tackle inequality instead of preserving the initial rank in 72% of the choices, and when preserving the initial rank involved slight increases in inequality between A and B, participants preferred to revert the initial rank 78% of times rather than increase inequality. In turn, the preference to allocate money to the participant with the higher performance in the effortful task was stronger than both the RRA (80% chose to revert the initial rank when relative performance mismatched the initial hierarchy) and the preference for reducing inequality (71% chose fairness even if it involved increasing inequality, and 67% chose fairness when pitted against eliminating inequality).

Conclusions. The present experiment presents a conceptual replication of the RRA first reported by Xie et al. (2017). Our original contribution is twofold: first, we managed to replicate the RRA with a methodologically cleaner protocol; and, second, we tested the relative strength of the RRA when pitted against conflicting values, in particular equality and fairness.