Data-Driven Differentiated Biology Instruction in a High-Stakes Secondary Classroom: An Action Research Study of Grade Progressions from Trial Examinations to SPM 2025

HRMARS - Improving Biology achievement in high-stakes secondary examinations requires pedagogical decisions that are both responsive to student readiness and defensible through evidence. This action research study examined the achievement trajectory of 147 Form Five Biology students across three assessment points: Trial SPM SBP, Trial SPM JPNS, and SPM 2025. The practitioner intervention was conceptualised as a continuous differentiated Biology improvement cycle involving diagnostic grade profiling, tiered remediation, targeted feedback, active learning, peer explanation, and repeated verification through assessment evidence. Supplementary qualitative evidence from teacher reflective notes and informal student exit slips was used to interpret possible mechanisms of change. Because grades were ordinal, each grade was converted into a grade quality point (GPK) using A+ = 0, A = 1, A? = 2, B+ = 3, B = 4, C+ = 5, C = 6, D = 7, E = 8, and G = 9; lower scores therefore indicated stronger achievement. Descriptive statistics, bootstrapped confidence intervals, Friedman’s repeated-measures test, Kendall’s W, Wilcoxon signed-rank tests with Holm adjustment, class-level comparisons, and baseline risk-group analyses were used. Mean GPK improved, decreasing from 5.97 in Trial SBP to 4.51 in Trial JPNS and 2.39 in SPM 2025. A-range grades increased from 7.5% to 59.9%, while pass attainment increased from 81.0% to 100.0%. Friedman’s test indicated a statistically significant and practically large shift, ?²_F(2, N = 147) = 223.67, p < .001, Kendall’s W = 0.761. All 147 students improved between Trial SBP and SPM 2025, with a mean improvement of 3.58 GPK points, 95% bootstrap CI [3.35, 3.81]. The findings suggest that a structured, data-guided, differentiated action research cycle can support substantial Biology performance gains, although causal interpretation is limited by the absence of a control group and lesson-level fidelity records.