Scoring and keying multiple choice tests : A case study in irrationality

We offer a case-study in irrationality, showing that even in a high stakes context, intelligent andwell trained professionalsmay adopt dominated practices. In multiple-choice tests one cannot distinguish lucky guesses from answers based on knowledge. Test-makers have dealt with this problem by lowering the incentive to guess, through penalizing errors (called formula scoring), and by eliminating various cues for outperforming random guessing (e.g., a preponderance of correct answers in middle positions), through key balancing. These policies, though widespread and intuitively appealing, are in fact ‘‘irrational’’, and are dominated by alternative solutions. Number-right scoring is superior to formula scoring, and key randomization is superior to key balancing. We suggest that these policies have persisted since all stake-holders – test-makers, test-takers and test-coaches – share the same faulty intuitions.