All of these methods focus on the control of over exposure, and it is generally assumed that underexposure problems will reduce when the exposure of the most informative items is limited. In practice, exposure control methods only increased the use of a subset of the under exposed item. Revuelta e Ponsoda (1998) addressed this issue. They proposed the progressive or restricted method for item selection where item selection is based both on the information provided by the item and on a randomized component. During the early stages of CAT the randomized component is more important. Towards the end, the information component is more important.