GENERALIZABILITY THEORY AND A SAMPLE APPLICATION FOR INTER-RATER RELIABILITY

Dr. Hakan Atılgan
Ege University

Abstract

The aim of this study is to introduce the G-theory, its essentials, and advantages over the Classical Test Theory. The G-theory, a flexible alternative for various measurement cases, has a very strong statistical base, and is an approach allowing determining reliability of measurement by taking into account all potential sources of error. Therefore the use of it by obtaining Phi coefficients for criterion-referenced testing, and G coefficients for norm-referenced testing has been demonstrated through a hypothetical example, in which raters get involved with measurement process, as a measurement case possessing various and multiple sources of error.

Keywords

Generalizability theory, inter-rater reliability, measurement, evaluation.