Accounts and first impressions in face-to-face and digital selection interviews using algorithms.
In the context of personnel recruitment, a structured selection interview is a reliable indicator of a person’s performance in the workplace. During a structured interview, recruiters often employ so-called behavioral questions, which prompt the candidate to provide a personal account concerning a past work situation. In addition to the fact that a candidate’s narrative responses are often suboptimal, it is difficult to assess whether their responses reflect a specific competence or if they stem from a general ability. Furthermore, with the arrival of new technologies, it is now possible to conduct digital selection interviews. However, candidates perform worse during digital interviews than during face-to-face interviews. Moreover, the validity of digital selection interviews has not been proven.
This project has three objectives. The first objective is to understand the extent that the performance of candidates’ narratives reflect on their expertise in a specific or general competence. The second objective is to understand the processes involved in the difference between performances during digital and face-to-face selection interviews. The third objective is to develop a verbal pipeline for digital selection interviews by improving automatic speech recognition and natural language processing.
This four-year project, which started in February 2021, is funded by the Fonds national Suisse with 700,325 CHF and involves Prof. Adrian Bangerter member of the LexTech Institute (Digital Economy Lab), Prof. Marianne Schmid Mast (University of Lausanne) and Phil Garner (IDIAP).