PRESENTER: Angela Verschoor, Senior Research Scientist at Cito, the Dutch National Institute for Educational Measurement.
The purpose of the theory test is to classify the candidates in two groups: those having sufficient knowledge on the theoretical aspects of traffic, such as the regulations, and those who fail to have this knowledge. On first sight, this classification sems to be quite straight forward, yet two things threaten this process: first, some candidates may have illegal access to the test questions, and second, misclassifications occur due to insufficient test quality as expressed in the test reliability.
Unfortunately, experiments already performed in the 1930s show that in regular school education, a low reliability is the greatest source of misclassifications of the two. Only situations with a massive breach in security is an exception to this observation.
In this presentation, the concept of reliability is explained as well as how to develop tests with high reliability. Security remains, of course, an important issue. This can be achieved by making sure that the composition of the test questions remains unpredictable for the candidates. Highly reliable tests that are unpredictable are not easy to develop. It is clear that a large pool of test questions should be available, preferably with metadata collected over previous test administrations.
Several methods for assembly of tests from such a large pool are discussed: random selection according to the specifications of the test, Automated Test Assembly (ATA) in order to construct several parallel test forms, and Linear On-the-Fly Testing (LOFT) are the most interesting approaches.
FURTHER DETAILS OF INTEREST FOR PARTICIPANTS:
- Time: 20 January 2022, 12:00 pm - 13:00 pm CET
- Fee: This members-only event is free-of-charge
- Location: online
- Working language: English
- Registration: The deadline for registrations is 18 January 2022