Should students change their answers on multiple choice questions? Believe it or not, that question has been explored empirically rather at length. Is it an important enough query to merit quantitative analysis? In and of itself, maybe not, but I wonder whether it isn’t related to questions with larger implications: How do you know if you’ve got something right? What makes you doubt an answer, and what questions should you ask when you do? What guidelines should you follow when guessing at the answer?