Abstract
The study analyzed the item parameter estimates of the Mathematics multiple-choice items. It also determined the reliability of the items, as well as the item characteristic curve of the items. These were with a view to providing empirical statistical evidence
on the nature of the items used by National Examinations Council. An ex-post-facto research design was adopted for the study. Senior secondary three students who sat for SSCE Mathematics examination in Osun State during the 2012 constituted population for the study. A sample of 2500 was selected through stratified sampling technique, using school location and sex as stratum. Data collected was based on responses to 60 Mathematics multiple-choice as contained in the scanned Optical Mark Record (OMR) sheets. Analysis of the data was carried out X-Calibre 4.2. The results of the item difficulty revealed that 16(26.7%), 22(36.7%) and 22(36.7%) of
the items were easy, moderately, and difficult respectively; in terms of discrimination indices, 23(38.3%), 13(21.7%), and 24(40.0%) had poor, marginal, and moderate discriminations respectively. The result also revealed reliability coefficient of 0.85 for the calibrated 60 Mathematics multiple-choice items, which implied appropriate internal consistency. The results finally revealed that 32 (55%) of the 60 items complied with the assumption of IRT with reference to Item Characteristics Curve under the 2-parameter model. The study therefore concluded that the Mathematics multiple-choice items used by NECO were moderate in terms of quality