Sourcing algorithms: Rethinking fairness in hiring in the era of algorithmic recruitment

Abstract Sourcing algorithms are technologies used in online platforms to identify, screen, and inform potential applicants about job openings. The popularity of such technologies is rapidly increasing due to their pervasiveness in online advertising and beliefs that sourcing algorithms can decrease time to hire while improving the quality of new hires. What is little known, however, are their potential risks: sourcing algorithms could (intentionally or unintentionally) encode or exacerbate occupational demographic disparities, thereby hindering organizational diversity and/or decreasing the effectiveness of online hiring practices. Because sourcing algorithms identify and screen potential job applicants before they are made aware of employment opportunities, methods for evaluating discrimination in hiring which focus solely on job applicants (e.g., adverse impact ratio), may fail to detect the effects of discriminatory sourcing algorithms. Thus, we propose an expanded model of the employee hiring process to take into account the role of sourcing algorithms. Based on empirical approximations, we conducted a Monte Carlo simulation study to examine the magnitude and nature of sourcing algorithms' influence on hiring outcomes. Our findings suggest that sourcing algorithms could hinder the diversity of new hires while misleadingly suggesting positive diversity outcomes in personnel selection. We provide practical guidance for the use of sourcing algorithms and call for a systematic re‐examination of how to evaluate selection system fairness in the era of algorithmic recruitment.