การเปรียบเทียบค่าความเที่ยงของแบบทดสอบผลสัมฤทธิ์ทางการเรียนที่มีจำนวนข้อสอบทำหน้าที่ต่างกันแตกต่างกัน

Abstract

A Comparison of Achievement Test Reliabilities under Increasing Levels of DIF

The purpose of this study was to compare reliability figures for test having a varying number of items with DIF, differential item functioning. Seven categories were used: test with no DIF items (0% DIF), tests with 5 % DIF, 10 % DIF, 15 % DIF, 20 % DIF, 25 % DIF, and 30 % DIF. The sample involved 2,000 Grade 6 students who sat a 40-item national achievement test on the Thai Language in a school under the jurisdiction of the Office of the Basic Education Commission. Descriptive statistics were derived by using SPSS; DIF was detected by use of the SIBTEST program. Gender was the classifying variable. The Z-test was used to compare test Reliabilities. The result was as follow: No significant differences were found among the Reliability coefficients as the number of DIF items changed over the seven levels mentioned.