EVALUATION OF DIFFERENTIAL ITEM FUNCTIONING (DIF) IN 2017 WEST AFRICANEXAMINATIONS COUNCIL (WAEC) PHYSICS MULTIPLE CHOICE TEST IN OSUN STATE,NIGERIA

Abstract

The study established if the 2017 WAEC Physics multiple-choice items satisfied of unidimensionality and local
independence assumptions of Item Response Theory (IRT). It also identified Physics test items that functioned
differentially for male and female testees as well as items that functioned differentially among urban and rural
examinees. The study adopted the descriptive survey research design. A sample of 720 senior secondary school
III Physics students was used for the study using the multi-stage sampling procedure in Osun State. The
research instrument used for data collection was an adopted 2017 WAEC’s Physics multiple choice questions.
The instrument showed a reliability coefficient of 0.85. Data collected were analysed using the IRT package of
R language and environment for statistical computing and Yen Q3 statistic. Results showed that 2017 WAEC
Physics multiple-choice items were essentially unidimensional (maximum DETECT value = -0.30 (< .20), ASSI
= -0.154 (< 0.25), and RATIO = -0.206 (< 0.36). Furthermore, none of the items violated the assumption of
item local independence of IRT since the residual correlation coefficients among pairs of items were less than
or equal to 0.2 as the benchmark. Finally, results showed that 12 items functioned differentially among male
and female examinees (p< 0.05) while 20 items functioned differentially among urban and rural examinees at
(p< 0.05). The study concluded that 2017 WAEC’s Physics multiple choice questions possessed items that
functioned differentially concerning gender and school location. The study recommended that DIF analysis of
test items should be done and that items identified to be functioning differently should be revised, modified, or
eliminated from the test before final administration to the examinees.