Valid, Efficient, Generative: iTest Reshapes Workplace English Assessment in University EFL Tracks

Abstract

In follow-up College English courses for non-English majors—specifically the Workplace English—the perennial headaches are heavy item-writing and spiraling costs. To address them, this study piloted a dual-cycle “stage + final” iTest workflow at Mianyang Teachers’ College (2024–2025). Using a convergent mixed-methods design, 116 non-English majors took adaptive CAT/LOFT exams while 124 peers sat equivalent paper tests. Results show 100 % syllabus-aligned item-bank coverage, item-writing time slashed 72 % (6.5 → 1.8 h), grading time cut 96 % (12.3 → 0.5 h), and direct cost per learner down 73 % (7.8 → 2.1 CNY). Final-exam reliability rose from α = 0.82 to α = 0.91 (p < 0.01); WEASE-Q satisfaction increased 18.6 % (5-point Likert) and test anxiety neared zero. Interviews confirmed “workload relief, data visibility, and track adaptability.” Findings position iTest as a scalable, psychometrically robust, and cost-efficient engine for digital ESP assessment in non-major programs.