Adaptive Testing for programming logic skills assessment

Adaptive Testing (AT) enhances learning outcomes by adjusting assessments to students’ proficiency levels. This paper presents adaptive methods for evaluating programming logic skills, implemented in an open-source system named MCTest. In this system, teachers create ATs tailored for their students. Three adaptive methods were developed: Semi-AT (SAT), Weighted Probability of Correction (WPC), and Maximum Likelihood Estimation (MLE). Six tests were designed, including a non-adaptive baseline, with multiple-choice questions classified according to Bloom’s Taxonomy. These tests validated item calibrations using Item Response Theory. The method was applied in two classes with 72 students, and a final questionnaire with 17 respondents statistically confirmed its perceived effectiveness.