Assessing Students’ Behavior in Error Finding Programming Tests: An Eye-Tracking Based Approach

Programming debugging is one of the most challenge part in the programming course, which is currently assessed by the teachers manually. With the development of eye-tracking technology, the student’s cognitive process can be estimated and researched based on the eye movement data. However, most of the existing eye-tracking measurement in programming focus on the difference among different person group or different mission, and can not be directly utilized for programming assessment. In this paper, we propose an assessing framework for debugging, providing quantified eye movement measurement on given errorfinding tasks. We focus on the task of finding errors in the C-language source code. An eye-tracking based measurement system is implemented to segmenting the students’ programming process in error-finding tasks. By dividing the source code AoIs and assigning the relevant exam point AoIs, the eye movement on specific exam point can be measured and analyzed. A procedural evaluation scheme is proposed to analyze the details of the testing process, including every error-finding activity and every eye movement. By checking the eye movement data on the related blocks, we estimate the students’ sensitivity to the specified examination point, and produce a reference classification on his performance. Experiment results show that, compared with the results of traditional error-finding question (only absolute true or false), our assessment method provides procedural evaluations of students according to his behavioral characteristics.