Investigation of the Reliability of Teachers, Self and Peer Assessments at Primary School Level with Generalizability Theory


Creative Commons License

GÜRLEN E., BOZTUNÇ ÖZTÜRK N., Eminoglu E.

JOURNAL OF MEASUREMENT AND EVALUATION IN EDUCATION AND PSYCHOLOGY-EPOD, cilt.10, sa.4, ss.406-421, 2019 (ESCI) identifier

Özet

This study aims at determining the reliability coefficients of teacher, self and peer assessments carried out at primary school level. In line with this aim, an interdisciplinary approach is adopted, and the notion of helpfulness included within the scope of values education is addressed in connection with the practices followed in Turkish, social studies and music lessons. The study group consists of 30 students of the third graders from a public school in the city of Ankara In the light of the aim of the study, the Generalizability Theory is used for the data analysis. It is found out at the end of the study that the variance component estimated for the main effect of the student is the largest component of the total variance in all three lessons. When G and (1) coefficients are examined, reliability coefficients are found to be over .80 in music, and over .90 in Turkish and social studies. According to G-Facet analysis results, when teacher and peer assessments are excluded from the analysis, respectively, G and (1) coefficients have a decreasing tendency whereas these coefficients increase when self-assessment is excluded from the analysis. Especially in the music lesson, the reliability coefficients obtained by excluding teacher and peer assessments from the analysis are found to be around .60, which is a remarkable result.