Changing Multiple-Choice Answers

When Told Not to Do So One of the authors wanted to determine the effect of changing answers on multiple-choice tests.

She studied the tests given by another professor, who had told his students before their exams that if they had doubts about an answer they had written, they would be better off not changing their initial answer.

The author went through the exams to look for erasures, which indicate that the first choice was changed. In these tests, there is only one correct answer for each question. Do the data support the view that students should not change their initial choice of an answer?