USING CHATGPT TO DESIGN ENGLISH PRACTICE TESTS FOR THE NATIONAL HIGH SCHOOL GRADUATION EXAM IN VIETNAM

The 2025 National High School Graduation Exam in Vietnam introduces changes to the English test format with more diverse question types, which may lead to some difficulties for both teachers and students regarding teaching and learning documents for practice. This study explores the use of ChatGPT in generating English practice tests aligned with the new 2025 national high school graduation exam format and evaluates the quality of the generated exams in terms of reliability, validity, and difficulty. Conducted at two high schools in a northern province of Vietnam, the research involved 85 to 154 students in grade 12 and followed five stages: designing prompts, gathering expert feedback, revising prompts, implementing the test with students, and evaluating outcomes. The practice test design process was carried out through eight stages: creating a test matrix, developing prompt templates, testing and refining prompts, storing as a validated template, including topics from the 2018 syllabus, creating questions for each part, reviewing and refining each part, and compiling and formatting the final test. A combination of expert judgment and statistical analysis was employed to evaluate the quality of the test. Expert feedback assessed face validity and content validity through Likert questionnaires. Students’ practice test results were analyzed using SPSS to examine construct validity, reliability (Cronbach’s alpha), and difficulty of items, tests. Findings show that with refined prompts, ChatGPT can create reasonably valid, reliable, and difficult practice tests by using the 8 stages of practice test design.<p> </p><p><strong> Article visualizations:</strong></p><p><img src="/-counters-/soc/0942/a.php" alt="Hit counter" /></p>