IRT Models for Subjective Weights of Options of Multiple Choice Questions

In the present models this approach is generalized by having the subject distribute a number of 'taws' over the options, or draw a line after the options, such that the number of taws given to an option, or the line length reflects its subjective degree of correctness. It appears that even with values of the relevant parameters that seem modest, the information relative to binary scoring still is in excess of two. This means that with less than half the test length the same accuracy or reliability can be obtained as with binary scoring. If a few main fallacies can be reflected in the distractors of the items, the model can be applied to identify subjects with one of these fallacies.