Detecting Local Item Dependence between the Test Item Pairs Using Q3 Index

Abstract

The present study aimed at detecting local item dependence (LID) between the test item pairs using Q3 index; identifying the percentage of item pairs that appear LID under different levels of examinees ability, and identifying the relationship between two Item Response Theory (IRT) assumptions: Unidimensionality and Local Item Independence, according to the 2–Parameters Logistic Model (2-PLM). To achieve this, the researcher used data available about the Computer Placement Test prepared by Al al-Bayt University, which is used to measure the freshman students’ level when joining the (BA) program in 2007/2008. The test consisted of (50) items measure some computer skills. The sample of the study consisted of (1108) examinees, distributed into five different sessions on the same test. To analyze the collected data, the researcher used three statistical programs (SPSS, Bilog-MG3, LDID) respectively. The results of the study indicated that the percentage of item pairs that reveal LID, using Q3 index, estimated at about 0.135. The results also indicated that the percentage of item pairs that reveal LID, using Q3 index, increased with increased examinees ability level. Finally, the results indicated that the two item response theory (IRT) assumptions: Unidimensionality and Local Item Independence were equivalent assumptions.