Abstract
BACKGROUND: Assessing a software engineer's ability to solve algorithmic programming tasks has been an essential part of technical interviews at some of the most successful technology companies for several years now. We do not know to what extent individual characteristics, such as personality or programming experience, predict the performance in such tasks. Decision makers' unawareness of possible predictor variables has the potential to bias hiring decisions which can result in expensive false negatives as well as in the unintended exclusion of software engineers with actually desirable characteristics. METHODS: We conducted an exploratory quantitative study with 32 software engineering students to develop an empirical theory on which individual characteristics predict the performance in solving coding challenges. We developed our theory based on an established taxonomy framework by Gregor (2006). RESULTS: Our findings show that the better coding challenge solvers also have better exam grades and more programming experience. Furthermore, conscientious as well as sad software engineers performed worse in our study. We make the theory available in this paper for empirical testing. DISCUSSION: The theory raises awareness to the influence of individual characteristics on the outcome of technical interviews. Should the theory find empirical support in future studies, hiring costs could be reduced by selecting appropriate criteria for preselecting candidates for on-site interviews and potential bias in hiring decisions could be reduced by taking suitable measures.