On the Suitability of Real-Time Assessment of Programming Proficiency using Gaze Properties

Abstract

Evaluating programming proficiency has become more relevant as the demand for coding skills has increased. Current methods, such as questionnaires or interviews, are methods that lack intuition, flexibility, and real-time capabilities. In our work, we investigate eye gaze behavior as an estimate for skill assessment. Specifically, we conducted a study (N=14) using an eye tracker to analyze the participants' abilities to understand source code by presenting them with a series of programs. We evaluated their eye movements based on common eye tracking metrics and identified mutual task-solving strategies among the participants. While we cannot relate these indicators to programming proficiency directly, this study serves as an evaluation of real-time methods for evaluating programming proficiency.