Both formative and summative assessments may involve three-fold purposes: to rank performances of all participants as a norm-referencing task, to order test items by difficulty and make diagnostic inferences for intervention, and to adjudicate which respondents have attained performance of a specified criterion level (pass or distinction). The use of a weighted or unweighted total score (percentage) as reportable outcome is an unquestioned convention in statistics teaching and examinations, and elsewhere. Several issues arise: under what conditions is a sum convention adequate, and when is it deeply problematic? How can the test data disclose evidence against the use of the sum? What then can be done to offer paths to justifiable outcomes that can stand both ethical and legal scrutiny? The case is made for the importance of prior relative difficulty judgments recorded within the marking memo rubrics, and the use of appropriate statistical methods for ordinal data.
(Tuesday 15th, 15:45-17:15) In session C8C
Timothy Dunne (South Africa)