Abstract:
At the Year 12 level in Victoria, as in several other State systems, there is widespread recognition of the value of ensuring that final assessments are based on a mix of subject-based examinations and structured school assessments. In particular, school-based assessment is seen as a means of ensuring assessment of the broadest possible range of valued outcomes and of improving the validity of final assessments.
The use of school assessments within a high stakes environment, does, however, raise significant issues regarding the comparability of those assessments. In the past, these concerns were addressed largely through statistical moderation which ensured that the location and spread of school-based assessments within a given school were adjusted to the location and spread of the examination results of students within that school. With the introduction of the Victorian Certificate of Education (VCE), statistical moderation was abandoned in favour of a system of verification based on external checking of samples of student work. The system proved inefficient and following reports by Ball and Brown (1992) and Hill, Brown and Masters (1993), the Board of Studies, Victoria, adopted a new system for ensuring comparability of school-based assessments. This involves all students taking a General Achievement Test (GAT) mid-way through Year 12. Scores on sub-sections of this test are used to construct composite scores which best predict students' results on school-assessed tasks. A two-level regression model is used to identify unexpected patterns of results and these are followed up by external markers. Where the assessments of a school are within the expected range given their scores on the GAT, the school's assessments are automatically confirmed.
This paper describes the new system, presents results from applying the new procedures in 1994 and provides a general discussion of the challenges presented in endeavouring to ensure fair and authentic school assessment at the Year 12 level.
The use of school assessments within a high stakes environment, does, however, raise significant issues regarding the comparability of those assessments. In the past, these concerns were addressed largely through statistical moderation which ensured that the location and spread of school-based assessments within a given school were adjusted to the location and spread of the examination results of students within that school. With the introduction of the Victorian Certificate of Education (VCE), statistical moderation was abandoned in favour of a system of verification based on external checking of samples of student work. The system proved inefficient and following reports by Ball and Brown (1992) and Hill, Brown and Masters (1993), the Board of Studies, Victoria, adopted a new system for ensuring comparability of school-based assessments. This involves all students taking a General Achievement Test (GAT) mid-way through Year 12. Scores on sub-sections of this test are used to construct composite scores which best predict students' results on school-assessed tasks. A two-level regression model is used to identify unexpected patterns of results and these are followed up by external markers. Where the assessments of a school are within the expected range given their scores on the GAT, the school's assessments are automatically confirmed.
This paper describes the new system, presents results from applying the new procedures in 1994 and provides a general discussion of the challenges presented in endeavouring to ensure fair and authentic school assessment at the Year 12 level.