On the quantitative analysis of assessment scores with implicit and explicit constraints

Abstract

Overrating or underrating assessment scores is common in educational settings. Solutions include rubrics, marker training, and moderation. A less common approach is score standardisation due to perceived complexity and challenges in maintaining the integrity of the original data. This paper presents a standardisation approach for assessment scores that transforms original data into final scores subject to implicit and explicit constraints. Scores from a unit administered under similar marking standards and policies at two campuses is used for verification. Implicit constraints address anomalies like diverse assessors, leniency, harshness in marking, and design variations. Explicit constraints arise from institutional policies and practices. We propose an analytical expression for transforming raw to final scores that preserves raw data's integrity while varying the standard deviation to satisfy constraints. A data filtering algorithm is applied to remove redundant and null scores. Verification reveals that the approach produces a Z-statistical score of 1.63, demonstrating comparable distributions.