The stability of marker characteristics across tests of the same subject and across subjects.

This research investigates the stability of marker characteristics within a very short period of time for both tests on the same subject as well as tests on different subjects. It reports on the scoring of the scripts of the whole cohort of students that took three high stakes tests in 2003 in a European country: a Language test consisting of a Literacy and a Literature paper and a History test. The many-facets Rasch model was used to study marker severity and marking consistency and it was found that some markers had more stable characteristics than others. Although the stability of marker characteristics was generally weak, it was non-negligible (correlation indices as indicators of stability ranged up to 0.707). This study, however, is not absolutely accurate due to the small sample sizes employed and it can be added that more research is needed to reach definite results.