• Version
• Download 105
• File Size 125.23 KB
• File Count 1
• Create Date August 2, 2018
• Last Updated August 2, 2018

Differential Speededness in the Context of Randomized Item Positioning

This study examines differential speededness when items in a test are presented to examinees in a random order. In computerized testing, item order is often randomized for the purpose of security and minimization of positioning effect on item parameter estimation. However, randomized item positioning might lead to undesirable differential speededness. When a test starts with timeintense items, examinees are likely to spend more time on them and therefore less likely to reach items at the end. Examinees under such condition could be unfairly disadvantage compared to those who start with less time-intense items. Empirical data from a large scale computerized test was analyzed to investigate the degree of differential speededness. Evaluation methods included proportion of items unreached conditioned on ability and average conditional differences in not reaching (Lawrence, 1993). The results indicated that in a completely randomized positioning design, differential speededness could be a potential threat to test fairness. The degree of differential speededness was especially prominent for lower ability candidates. Possible remedies ,to the undesirable differential speededness were briefly discussed.

Attached Files