Investigating patterns of Rasch item estimates in examinations of educational statistics courses in University Of Cape Coast, Ghana: a gradient assumption examined

Abstract

From a theoretical lens, Rasch modeling has long been an approach for educational measurement. Nonetheless, it appears that the Rasch model has not been fully exploited within the Ghanaian Context. This paper investigates patterns of Rasch item estimates in examinations of educational statistics courses at the University of Cape Coast, Ghana. The population for the study was all levels, 100, 200, and 800 students who offer statistics courses. Purposive sampling was used to select 150 students (offering educational statistics courses that are; EPS 211, EPS 121, EPS 851, and EPS 856). The instrument for the data collection was the examination results in these four courses in educational statistics. The data were analyzed using item difficulty and ability parameter logits of the Winsteps Rasch Measurement Model computer program. The study used the ex-post facto design. The findings of the study showed that item difficulty parameters were appropriate but were not arranged hierarchically from the least difficult to the most difficult item and students' ability parameters were appropriately estimated by the examination items. It was recommended among others that, for valid and reliable items with appropriate item difficulty parameters and person ability parameters that measure the intended unidimensional construct educational statistics lecturers should consider using the Rasch Model in developing their examination items.