(15pp.) An increasingly popular test format allows examinees to choose which items they will answer from among a larger set. When examinee choice is allowed, fairness requires that the different test forms thus formed be equated for their possible differential difficulty. For this equating to be possible, we need to know how well examinees would have answered the items that they did not choose. In this paper we report the results of an experiment in which examinees are asked to choose among several multiple-choice items but are then required to answer all of them. We conclude that allowing choice while having fair tests is only pos- sible when it is unnecessary.
