A further trial of a bookmark approach to grading objective tests
By Diana Fowles
A second trial of the Bookmark method has been conducted, in the context of a GCSE Science module, on the basis of a one parameter model analysis of the objective test data. Both trials have shown variation in outcomes from different judges which might be explained by a lack of a shared understanding of what '67% probability of success' means at each grade, something that might be improved with practice and training.
For each grade the boundary marks selected in the awards on statistical grounds were closer to matching 67% than were the judges' Bookmark selections.
The information on item and test performance from the Rasch analyses might be more readily and effectively put to use than in the production for the Bookmark method of the ordered booklet and the judges' selection of the Bookmark item.
A variation on the Bookmark task is suggested in which the '67%' item is identified for every possible boundary mark (this is the item on which candidates with that total mark have closest to 67% probability of success according to the model) and offered to the judges (in the form of a table) for their selection of the best one to represent performance at the particular judgemental grade boundary.