Document Type : Research Paper

Authors

• Reyhane Rahimi ¹
• Aso Mojtahedi ²

¹ Assistant Professor, Department of Educational Science, Farhangian University, Tehran, Iran

² Ph.D. Student in Measurement & Assessment, Allameh Tabataba'i University, Tehran, Iran

Abstract

 
The goal of this research is to explore the Likert scale questions using two distinct methods: Classical Test Theory and Item Response Theory. By comparing the results of these approaches, the study aims to address the question: "Do the outcomes from these two methodologies align, or do they contradict each other?" The research design followed a descriptive methodology and utilized secondary analysis techniques. The study population consisted of 977 junior high school students. After the data screening process, the final sample size for analyzing extraversion items was 783 students, 763 students for openness items, and 784 students for conscientiousness items. The research instruments were the three subscales of extraversion, openness, and conscientiousness from the Neo Personality Test. The statistical analysis yielded results indicating that a strong internal consistency among items enhanced the accuracy and validity of outcomes derived from the graded response model. However, when items exhibit low internal consistency, caution should be exercised, as the model may yield erroneous thresholds or discrimination coefficients (i.e., false negative or positive). Overall, combining multiple methods of statistical analysis can significantly contribute to more effective analysis and obtaining highly accurate results.

Keywords

• Graded Response Model
• Item Response Theory
• Classical Test Theory
• Internal Consistency
• Threshold
• Discrimination Coefficient