Examining Item Position Effect on Reliability Coefficients of Multiple Choice Physics Achievement Test

This study explored the effect of changes in item sequence on test reliability coefficients (test-retest and Kuder Richardson 20) and student’s achievement in multiple-choice physics tests in Senior Secondary School II in Taraba State. The study adopted repeated measures two-group within-subject experimental research design. The research collected data in order to answer three research questions and test one hypothesis. The sample comprised 450 senior secondary II Physics students (male = 303 and female = 147 ) drawn from population of Physics students in Taraba State. Multi stage sampling technique was employed to randomly select twelve schools from three Local Government Areas of the three Senatorial Districts of Taraba State, and an intact arm of SS II from each of the sampled schools was used. Two parallel 40-items Physics Achievement Test developed by the researcher were used for data collection. The resulting data were collated and analysed using descriptive statistics and t-test. Results showed that reliability coefficients obtained from the two formats were quite comparable. However, the finding revealed that there is significant difference between the students mean achievement score in the format A and Format B of the physics achievement test (t = 4.409 , df = 898, p < 0.05, two-tailed). Sequels to the findings, it is concluded that changes in item sequence has no effect on the reliability coefficients, but students will perform better in physics achievement test if the test items are arranged randomly than in descending order of difficulty.