Aversion to hiring algorithms: Transparency, gender profiling, and self-confidence

How do these properties interact? Severe consequence: Listening for 10 minutes to a painful tone (85 dB and 2083 Hz) Harmless consequence: Waiting for 10 minutes Lab experiment • Exchange of 10 minutes of painful tone [waiting time] against money • Before any offer can be made, spectators decide whether transaction may take place or not • All participants listen to the tone for one minute before t. [...] Lab experiment 10 € 10 € 10 min 10 min A B Lab experiment 10 € 10 € 10 min 10 min X € 10 min A B Lab experiment 10 € 10 € 10 min 10 min X € 10 min accept or reject A B Limited Agency Full agency B can accept or reject offer. [...] Conclusions: Repugnance • Some goods affect who we are and what society we live in, and some goods save lives • Limiting market transactions for those goods because of repugnance can be hard on those who want to engage in them • How can we organize the allocation of such goods in a manner that makes the transactions acceptable? - Ensuring agency, e.g., by consent procedures, poverty reduction, pub. [...] Conclusions: How to avoid black markets • First-come first-served systems are vulnerable to scalping • Proposed batch system also deals with complications (such as scalper submitting applications with fake IDs) • Similar to re-sale of tickets for concerts and sporting events, limited- edition sneakers etc., but appointments are for free • Some tourist sites use similar procedures as batch system b. [...] • In BaselineM and TranspM, overconfidence is negatively correlated with delegation • Higher overconfidence is associated with a significantly stronger treatment effect of ConfidM Performance and gender • In task 1 (Raven matrices), men perform better than women • In task 2 (counting zeros) and in job task 3, no significant gender difference • For the same task-1 and task-2 performance, women perf.