Abstract:

This study compares student behavior when answering multiple-choice questions based on two complementary experiences conducted at different academic levels: undergraduate students assessed using traditional single-answer tests, and master’s students evaluated through Confidence-Based Marking (CBM). The aim is to examine how response strategies, metacognitive engagement, and the quality of evaluative information differ depending on the test format and the learners’ academic stage.

In traditional tests administered to undergraduate students, two opposite patterns frequently emerge: on the one hand, a tendency to guess when uncertain; on the other, a substantial number of unanswered questions. This widespread omission makes it difficult to determine whether students skip questions due to lack of knowledge, low confidence, poor exam strategies, or a combination of these factors. As a result, the evaluator’s ability to accurately interpret the students’ actual level of mastery is significantly limited.

In contrast, master’s students assessed using CBM exhibit more self-regulated behavior: the requirement to explicitly declare confidence levels reduces impulsive responding, minimizes unanswered items, and allows a more accurate calibration between perceived certainty and actual probability of correctness. Furthermore, the CBM format provides clear indicators of overconfidence, justified uncertainty, and potential misconceptions, offering a richer understanding of the cognitive processes underlying each response.

Overall, the findings suggest that incorporating confidence levels enhances the diagnostic precision of assessment and promotes metacognitive reflection, particularly in advanced educational contexts. At the same time, the results highlight the limitations inherent to traditional tests—especially when high omission rates occur—in understanding student behavior. Taken together, the study underscores the potential of CBM to complement or transform current assessment practices and to provide more robust insights into learners’ knowledge and reasoning.

Keywords:

CBM, test, exams, evaluation.