Better job application systems: Objectively assessing measures of job performance from asynchronous video interviews

When selecting top candidates for a job, organizations would prefer to not accidentally filter out the highest quality candidates. But an unbiased, detailed assessment of every applicant in a large candidate pool has been prohibitively costly. Asynchronous video interviews (AVIs) are inspiring new ideas for predicting job performance early in the hiring process by providing a rich source of signals. We propose that automatic analysis of interview data can improve candidate filtering at the application stage. The potential of this approach is clear, but there is a need for a structured framework and benchmarks to develop effective and valid application systems. We therefore propose a design framework that enhances the objectiveness of automated candidate assessment using AVIs through principles such as using behavioral cues that are hard to fake and using unbiased, validated labels in training sets. We demonstrate the implementation of this framework and evaluate its potential by building a prototype for automatically assessing general mental ability, an important and generalizable indicator of job performance. Results show that if new application systems adhere to this framework, more objective measures of job performance can be assessed automatically from AVI recordings. More generally, the study guides advancement of automated AVI platforms with a focus on efficacy and fairness.