Trend Analysis of Artificial Intelligence-Generated Econometrics Multiple Choice Test Item Parameters Among Undergraduate Economics Students, Nasarawa State University, Keffi, Nigeria

This study examined the psychometric characteristics of Artificial Intelligence (AI)-generated Econometrics multiple-choice test items Parameters (EMCTIP) among undergraduate Economics students, Nasarawa State University, Keffi, Nigeria. The study specifically determined the trend in the difficulty indices, discrimination indices, and effectiveness of distractors of AI-generated test items. A descriptive evaluative research design was adopted. The sample comprised 400 undergraduate Economics students. A 30-items AI-generated Econometrics multiple-choice test (EMCT). The EMCT was subjected to experts’ judgement, and yielded validity index of 0.74, and 0.073 reliability index. Item analysis was conducted using difficulty index, discrimination index, and distractor effectiveness as the major indices. Means and standard deviations was used to answered the psychometric indices and examine their trends across successive test administrations and One Way ANOVA to test the hypotheses at 95% confidence level. The findings revealed that there is a significant difference in the difficulty indices of AI-generated EMCT items; there is no significant difference in the effectiveness of distractors of the AI-generated EMCT items; and there is no significant difference in the discrimination indices and effectiveness of distractors of AI-generated EMCT items across successive test administrations among undergraduate Economics students of NSUK, Nigeria. The study concluded that AI-generated EMCT items demonstrated satisfactory psychometric properties in terms of difficulty, discrimination, and distractor effectiveness. It was recommended that lecturers and test developers should subject AI-generated test items to systematic human review, empirical item analysis, and validation before their use in formal examinations.