What is happening when we assess, and how can we use our understanding of this to improve assessment?

In an attempt to gain a fuller understanding of the basis of grading, ten assessors each assessed two portfolios drawn from the course archive which had been produced by participants on a course in teaching in higher education. Assessors gave a grade or judgement on each of a portfolio's 75 portfolio elements, reasons for each judgement they made, and recorded any alternative judgement they had considered. There were substantial differences between the judgements made during the study and those made when the portfolios were originally assessed. This article explores the differences between the original and experimental assessments, and then analyses reasons given by assessors for the judgements made. Beyond these particulars, the paper suggests some fundamental and problematic issues about the conduct and the reform of assessment, and offers a process which can be used to investigate and improve the quality, in particular the reliability, of any summative assessment.