The study aimed to conduct a Rasch analysis of the Edward Personality Preferences Schedule (EPPS) to evaluate and enhance the instrument's precision. Responses of 3.125 students in West Java, Indonesia, were subjected to Rasch analysis using the partial credit Rasch model. The Initial analysis supported the use of the Rasch model, and acceptable reliability (person separation index= 0.82) was observed. By creating super-items, a local dependency between items increased model fit, reliability (person separation index =0.85), and unidimensionality. Infit and outfit statistics for all items satisfied the criterion for construct validity. Almost all items except (17, 97, 126) had suitable point-measurement correlation, reflecting content validity. Item characteristic curves indicated that What evenly distributed roughly 225 items along the person ability continuum. The modified EPPS showed excellent reliability for a measure in assessing the overall potential ability of the students. The item would be added related to the person ability estimates. And Who should omit several items to ensure construct validity.
Copyrights © 2021