Implementing (Un)fair Procedures: Containing Favoritism When Unequal Outcomes are Inevitable [Dataset]

We study the selection of people when unequal payoffs are inevitable, but fair procedures are feasible, as in selecting one person from several candidates for a job. We show that allocators may be influenced by their similarity with a recipi- ent, leading to favoritism in outcomes. We study four interventions to reduce favoritism and induce fair procedures, without restricting the allocator’s decisions: transparency of the allocation process; a private randomization device; allowing the allocator to delegate to a public randomization device; and allowing the allocator to avoid information about recipients. Making use of beliefs and fairness judgments, we show why some interventions work, while others do not.