Evaluating a stratified College English teaching reform using student-level CET-4 data: a quasi-experimental study in China

Background Learner heterogeneity remains a persistent challenge in College English teaching in Chinese higher education, especially in health-related institutions where students often enter with markedly different English foundations, learning pace, and support needs. In September 2023, our institution introduced a placement-based stratified teaching reform across four colleges, grouping first-year students into A, B, and C levels according to placement-test performance. This study evaluated whether the reform was associated with improved College English Test Band 4 (CET-4) outcomes and whether the placement mechanism showed predictive validity for subsequent performance. Methods We used a two-stage quasi-experimental design. Stage 1 was a repeated cross-sectional analysis of student-level CET-4 records from June 2021 to December 2025. Primary outcomes were CET-4 total score and pass status ( > = 425). Multivariable linear and logistic regression models compared pre-reform (2021–2022) and post-reform (2024–2025) outcomes, adjusting for college and examination month, with cluster-robust standard errors at the student level. The year 2023 was treated as a transition period because the reform began in September 2023 and therefore did not represent either a fully pre-reform or fully post-reform instructional environment. Stage 2 was a retrospective linked cohort analysis linking placement-test data from the 2023 and 2024 cohorts to subsequent CET-4 records to examine the predictive validity of placement level and baseline placement score. Results A total of 28,160 CET-4 records from 13,233 unique students were included. Across examination cycles, pass rates were generally lower in 2021–2022 and higher in 2024–2025. After adjustment for college and examination month, the post-reform period was associated with higher odds of passing CET-4 (OR 2.13, 95% CI 1.97–2.30; P < 0.001) and a 43.45-point higher total score (95% CI 40.14–46.76; P < 0.001). Improvement was observed across all four colleges, although the magnitude varied. In the linked cohort analysis (n = 4,983), clear gradients were observed across A/B/C levels. Compared with C-level students, B-level students had higher odds of passing CET-4 (OR 3.09, 95% CI 2.55–3.76), while A-level students had markedly higher odds (OR 11.00, 95% CI 8.90-13.59). Placement score was also positively associated with later passing probability (OR 1.084 per point, 95% CI 1.077–1.091). Conclusions The 2023 placement-based stratified College English teaching reform was associated with improved CET-4 outcomes at the institutional level. In addition, the A/B/C placement system demonstrated clear predictive validity for subsequent CET-4 performance. These findings support placement-based differentiated teaching as a feasible institutional strategy for addressing learner heterogeneity in College English education within health-related higher education contexts.