| Abstrakt: |
In foreign language testing, reliable measurement of writing abilities has often been challenging. It is claimed that language schools would better have their current scoring standards tested regularly. In case low reliability levels are taken, to design their own criteria according to their goals would be better for such schools. In this study, it was aimed to find out the reliability levels of the holistic instrument that was used at Anadolu University. With this aim in mind, a total of 50 papers of different achievement levels were selected and given to ten graders who have a minimum of 3-year-experience in grading. These graders were asked to grade these papers using their current criterion twice with 1-month interval. Both the inter/intra-rater reliability levels were computed and a need to develop a new criterion emerged after it was found that the reliability levels were considerably low. With the new criterion designed to measure analytically, the same papers were again graded by the same graders. To confirm those calculations, the same papers were graded with each criterion by the same rd graders for the 3 time after six months, and the results of the reliability tests were compared for a final analysis. [ABSTRACT FROM AUTHOR] |
