Data driven job matching has become increasingly popular in organizations in recent years. Organizations use matching algorithms for matching people with jobs for example in recruitment and selection, job rotation, and during reorganizations. Research in work and organizational psychology shows that a good match between people and their jobs (i.e., person-job fit) has several benefits for both employees and organizations. However, for determining a match, research on person-job fit has relied mostly on one employee-perspective-based specific criterion and it does not sufficiently take into account the complexity of person-job fit.
In this PhD project, we will first focus on data driven job matching using matching algorithms, and examine the consequences on employee performance and attitudes at work. Thus the first aim is to get better insight into the complexity of person-job fit, and better understand the set of factors that determine a (successful) job match. Second, we will focus on comparing algorithms for job matching problems, which are not only matching algorithms that use a limited set of criteria, but also the recent machine learning algorithms that can learn from datasets with many features. On the other hand, since these algorithms are difficult to interpret, despite their accurate results, there is a growing interest in interpreting them and designing highly accurate yet interpretable new machine learning methods. Thus the second aim of this project is to compare existing matching algorithms against several interpretable machine learning algorithms.