Assessment of Psychometric Qualities of Gender Performance in Basic Education Certificate Examination in Mathematics Multiple Choice Test Items
Keywords:
Assessment, Calibrate, DIF, Item Bank, Rasch, IRT, 3-PL ModelAbstract
The study was aimed at assessing the psychometric qualities of gender in
Mathematics multiple choice test items in Basic Education Certificate
Examination (BECE) in Oyo State. Four research questions and one hypothesis
guided the study and was tested at 0.05 level of significance. An instrumentation
research design was adopted. The population of this study consisted of all Junior
Secondary class three students in Ibadan, Oyo State. The multistage sampling
technique was employed to sample out 750 students from the population. The
instrument used for data collection was the 2017 BECE Mathematics multiple
choice test items. The reliability value of the items using KR20 was 0.78. DIF
measure statistics of Winsteps was used to assess the difficulty indices of both
the male and the female involved in the study and also tested the stated
hypothesis at 0.05 level of significance. The finding revealed that there were
incidences of gender variance in the test scores. Reliability with IRT of .98
showed high representativeness of the items but lack local independence of item,
hence, it lacked unidimensionality. The test contained 19 bad items and 41 good
ones, therefore it was concluded that the 19 items should either be deleted,
reviewed, removed or restructured. It was therefore recommended that
examination bodies should be mindful of the existence of ‘item noise’ which could cause bias in gender performance in an examination. It was equally
recommended that government should developed calibrated MAT items Bank.