Answers are grounded in this paper and cite their sources.
This study presents a level-weighted scoring protocol for assessing English language listening and reading proficiency, integrating Computerized Adaptive Testing with an analytic grading formula in which incorrect answers at lower levels carry a greater penalty than errors at advanced levels. The methodology is evaluated via Monte Carlo simulation (10,000 iterations per profile) across four simulated student profiles: high-ability, low-ability, random, and inverse – the last representing a learner who fails foundational items while succeeding at advanced ones. The proposed formula correctly assigns a lower mean score to the inverse profile (1.36) than to the lowability profile (1.51), a discrimination the standard guessing-corrected system fails to achieve (4.18 vs. 3.38). It also suppresses random-guessing scores more effectively (mean 0.55 vs. 1.63). These results demonstrate that the score penalty applied to foundational-level errors improves assessment accuracy in normative profiles by enhancing the difference between students with gaps in basic knowledge and those with a clear mastery of essential content.
No takes yet. Share an insight, caveat, or question.
Martín-Campoy et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: