ASSESSMENT OF PSYCHOMETRIC PROPERTIES OF GOVERNMENT TESTS IN OSUN STATE UNIFIED PROMOTION EXAMINATIONS

Abstract

The dimensionality, item difficulty, and discrimination indices of test items used in the government tests of Osun State Senior Secondary Schools' Unified Promotion Examinations (UPE) were investigated in this study.

For this study, an ex-post facto design was used. The population of the study comprised 27, 470 Senior Secondary 2 (SS2) students from Osun State who took the 2019 Unified Promotion Examinations (UPE). Using a multistage sampling technique, the study sample consisted of 4,423 SS2 students who sat for Government test in the examinations. Using a simple random sample technique, nine Local Government Areas (LGAs) were chosen from the state's three Senatorial districts. Due to the presence of SS2 students in the schools, all Senior Secondary Schools in the LGAs were purposefully selected to participate in the study. All students who took Government were chosen using a purposeful sample technique. The data was analyzed with the sirt models package.

The Government examination was found to be unidimensional (Detect 0.20, Assi 0.25 and Ratio 0.36). The difficulty index  ranged between 50% and 51%, according to the findings. In addition, the discrimination index returned discrimination parameter estimates that were more than 1.7. Out of the 40 items tested, item 7 was found to violate item local independence. Finally, when the model-data fit was evaluated, 4PL was shown to have the best match (P< 0.05, RMSEA>0.05).

The study concluded that the multidimensional 4 parameter IRT model (M4PL) is reliable and acceptable for evaluating the psychometric quality of test questions used in the Osun State Unified Promotion Examinations Paper I Multiple Choice Examinations.

Dimensionality, Multidimensional Item Response Theory, Psychometric Property