Retesting Trajectories and Benchmark Attainment in the College English Test Band 4: Longitudinal Assessment Analytics From Six Administrations
Sage Open · 2026-10 · 16 (4)
DOI: 10.1177/21582440261504640 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sage Open, online erschienen am 13.11.2026. Verfasst von Yadan Wang und 3 weiteren Autor:innen. Die Arbeit nennt 36 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0). Gefördert von: Project of Shaanxi Federation of Social Sciences.
Abstract (Originalsprache)
Standardized English examinations are often summarized through administration-level means and benchmark rates, although repeated participation can create a more selective longitudinal record. This study analyzed six College English Test Band 4 (CET-4) administrations over three years at one private university in southern China. The six administrations were observation opportunities rather than six scheduled waves for every student: students who reached 425 could, if they had completed the required CET-6 coursework, become eligible to register for CET-6, making subsequent CET-4 participation potentially selective with respect to prior performance. After excluding zero-score query records and component-inconsistent reports, the analytic dataset contained 36,486 valid CET-4 records from 20,681 students. Analyses examined administration-level score distributions, normalized component profiles, observed score movement among 9,496 students with at least two retained CET-4 records, and adjacent next-attempt transitions among students whose prior score was below 425. The overall CET-4 mean was 398.89 (SD = 52.36), and 29.6% of valid records reached 425. Listening had the lowest normalized component mean. Among repeat test-takers, the mean first-to-last observed change was 21.03 points, 95% CI [20.16, 21.89], and 34.8% of initially below-425 retesters later reached 425. In group-based fivefold cross-validation, a component model provided only modest incremental discrimination over a prior-total-score model (AUC = .734 vs. .712; Brier score = .146 vs. .150). The findings describe selective movement through one institutional CET-4 testing system rather than causal learning growth and show what person-linked repeated records add beyond a conventional administration-level report.
Förderung
- Project of Shaanxi Federation of Social Sciences
Zitieren
APA
Wang, Y., Zhang, J., Sun, R., & Singh, S. S. B. (2026). Retesting Trajectories and Benchmark Attainment in the College English Test Band 4: Longitudinal Assessment Analytics From Six Administrations. Sage Open, 16(4). https://doi.org/10.1177/21582440261504640
Vancouver
Wang Y, Zhang J, Sun R, Singh SSB. Retesting Trajectories and Benchmark Attainment in the College English Test Band 4: Longitudinal Assessment Analytics From Six Administrations. Sage Open. 2026;16(4). doi:10.1177/21582440261504640
BibTeX
@article{Wang2026,
author = {Wang, Yadan and Zhang, Jin and Sun, Rong and Singh, Soon Singh Bikar},
title = {Retesting Trajectories and Benchmark Attainment in the College English Test Band 4: Longitudinal Assessment Analytics From Six Administrations},
journal = {Sage Open},
year = {2026},
volume = {16},
number = {4},
doi = {10.1177/21582440261504640}
}
EUVIMED