Diversity goals - whether explicit or implicit - are usually defined on a large scale, such as the student body of a university or the workforce of a company. However, their implementation often occurs at a smaller scale, as hiring and admission decisions are made by different decision-makers (e.g., HR managers, admission officers) at different points in time (e.g., recruitment rounds). We study the consequences of this phenomenon for selection outcomes, using register data from the selection process of a large study grant program. In the process, evaluators assess a limited number of randomly assigned candidates and vote on their admission outcomes. Our findings provide evidence that evaluators aim at balancing their small number of positive votes with respect to gender, migration background and socio-economic status. Consequently, individual admission chances decrease with the number of other candidates sharing the same attributes in the evaluator’s pool. Our findings highlight that the small-scale implementation of diversity efforts can distort selection outcomes, even when there is no quality-diversity tradeoff at the aggregate level of the candidate pool.
The Discussion Paper on an AI-related topic published in 2024 by Christina Gathmann (LISER and IZA), Felix Grimm (LISER), and Erwin Winkler (University of Erlangen-Nürnberg and IZA) has been selected for the 2025 IZA Award for Innovative Research on a Pressing Public Issue (IRPPI).
