Abstract
In a series of papers ov er the last ten years, I have outlined various problems affecting
the assessment of the national curriculum in England which are the subject of a critique
by Paul Newton (this issue). In responding to this critique, I acknowledge that his
summary of my position is fair, and agree that, by the standards of analytic rationality,
the evidence for some of the problems I identify is not compelling. However, in
response I argue that by standards of reasonableness (eg on the balance of probabilities)
the evidence is sufficently serious to warrant a re-examination of national curriculum
assessment, and the alternatives. In particular, I argue that the current system provides
assessments that are not sufficiently reliable for the inferences that are made on the
basis of the results and has also caused a narrowing of the curriculum. I propose that the
first of these weaknesses can be addressed through the increased use of teacher
assessment, and the second by increasing the range of the curriculum tested through
testing a greater proportion of the curriculum. In order to effect these changes without
increasing the burdern on students and teachers, I propose that these two changes are
combined in the form of a light sampling scheme which would increase both the
reliability and minimise the curricular backwash, although the price paid for this would
be the lack of a direct, transparent and objective link between the results achieved by
individual students on tests and the reported levels of a school’s performance.
Key words
education, national curriculum assessment, reliability, validity, manageability