Research Seminar
Modeling item order differences within a DIF frameworkMussie Kebede AkaluKU Leuven | |
| Abstract: | In a computerized test administration on listening ability in French, Flemish students from the second year of secondary education took one of two tests, which had a number of anchor items in common. The test consisted of a number of audio fragments, which were followed by one to three questions. For each test, the audio fragments were administered in one of two orders, with one order being the reverse of the other order. For both tests, the data showed a clear effect of item order: when items appeared in the last part of the test they were more difficult than when presented at the beginning. It is shown that this effect test fatigue can be modeled within a DIF framework. The advantage of this calibration strategy is that the item parameters can be estimated without the effect of item order included. |
| Date: | Tue Apr 24, 1:00 pm - 2:00 pm |
| Place: | room 00.60 (Department of Psychology, Tiensestraat 102, 3000 Leuven) |