Step 3. Administration
In the third step the item is presented to the candidate. The mode of presentation has to be robust against the use of various computer platforms and various types of monitors. It has to be guaranteed that every candidate can respond to the item based on the same amount of information without any distraction. After presenting the item, a candidate either has limited or unlimited time to answer the item. When response times are limited, one might consider the use of response time models (VAN DER LINDEN, 2007) to correct for speededness of the test. In some CATs, several items are presented on the same page, for example, when they are all related to the same stimulus. In most CATs however, only one item is presented at a time. It is important to realize that as a result of selecting each subsequent item based on information obtained in previous items, in most CATs it is not allowed to review earlier responses. Allowing item review would reduce measurement efficiency and would make CAT vulnerable to test taking strategies (WAINER, 1993), besides the assumption of local independence would be violated, that states that the observed responses are independent of each other given an individual’s score on the latent trait. Bowles e Pommerich (2001) studied the impact of item review and found only limited effects on bias and root mean squared error of the ability estimates