MEASURING THE RELIABILITY OF NATIONAL CURRICULUM ASSESSMENT

Conventional measures of reliability cannot easily be applied to Natio nal Curriculum assessment. There is a need to develop methods which te ll us how reliably pupils are allocated to Levels within each attainme nt target. The situation is complicated by the variety of different me thods used to award Levels based on marks obtained or performance on S tatements of Attainment. Statistical models can be used to develop gra phical displays of the uncertainty in Level assigned against pupil pot ential attainment, and study of these graphs gives important insights into the functioning of National Curriculum assessment.