ASSESSING THE INTERNAL STRUCTURE OF MATHEMATICS ACHIEVEMENT TEST FORJUNIOR SECONDARY SCHOOL STUDENTS IN OSUN STATE

Abstract

The study assessed the internal structure of mathematics achievement test. A descriptive survey research
was adopted for the study. The population consisted of all Junior Secondary School 3 (JSS 3) Students in
Osun state. The purposive sampling technique was used for the study because of the homogeneity
characteristics of interest to the researcher. Hence, Osun central senatorial district was selected for the
study. 5 Local Government areas were randomly selected from the 10 Local Government Areas in Osun
central senatorial district, from each of the 5 selected Local Government Areas, 2 schools were selected
from each of the Local Government area making total of 10 Junior Secondary School in Osun state. The
sample size comprised 600 students, from each of the 10 schools, an intact class was used. 60 students
OMR sheet were randomly selected from the intact class, from the selected schools given total of 600
students. The instrument for this study was (2012) multiple – choice in mathematics of Basic Education
Certificate Examination for Junior Secondary School in Osun State titled Mathematics Achievement Test
(MAT). The data collected were analyzed using Stout’s Test of Essential Unidimensinality (STEU) and
Correlation Residual with Yen Q3 Statistics. The result showed that more than one dimension was
accounted for based on the variation observed in examinees responses to the test items, it further revealed
that 48 (80%) out of 60 items of 2012 Multiple – Choice in Mathematics of Basic Education Certificate
Examination in Osun State violated the assumption of item Local Independence. The study recommended
that for quality items to be guaranteed, the test items should be subjected to unidimensional and local
independence assumptions of IRT frame work