Examining the differential rater functioning in the process of assessing writing skills of middle school 7th grade students

Abstract

When students present writing tasks that require higher order thinking skills to work, one of the most important problems is scoring these writing tasks objectively. The fact that raters give scores below or above their performance based on several environmental factors affects the consistency of the measurements. Inconsistencies in scoring negatively affect the validity and reliability of student performance and cause the scores obtained to be questioned. In regard to the validity and reliability of these measurements, it is significant to identify the rater behavior and correct the sources of error. This study aims to analyze the differential rater functioning (DRF), which is one of the problematic rater behaviors, in evaluating compositions written by middle school 7th-grade students within the scope of the Turkish course. 86 students attending a public school were participated the study. Students' compositions were rated using an analytical rubric by 8 teachers from different institutions. In this correlational research, the many facet Rasch model was used, and five variables including students, raters’ and, students’ gender, students’ qualification, and evaluation criteria were examined. it was examined whether the raters show DRF on an individual and group basis based on the dual interaction analysis, including the gender of the student x rater and the student's competence x rater. The findings have revealed that DRF at the group level does not interfere with the measurements, while the individual level DRF is involved in the measurements. It was determined that the level of DRF mixing in the measurements of successful students was the lowest. Especially rigid and lenient raters were found to show DRF. In the present study, it was observed that the raters showing DRF was also the most lenient raters, while these raters did not show DRF in terms of the gender of the student