Adaptive testing
The adaptive testing is particularly in computer-based psychological tests used. During the test, the test person is not asked all questions or tasks, but only those that correspond to his level. This enables sufficient information to be gathered in a shorter period of time to locate the subject with regard to the characteristic of interest. This increases the test economy. Item response theory (also known as probabilistic test theory ) is used to construct an adaptive test .
Theoretical background
The basic idea of adaptive testing is to only use the questions that provide the most information about how strong the trait of interest is in the subject. For example, an item that is too easy does not provide any information for the respective test person because this question would be answered positively by almost every test person with similar characteristics. Questions that are too difficult also have no information contribution, since these questions would not be answered by most of the test persons with similar characteristics. Only questions that have an average task difficulty for the test person are clear-cut, ie able to differentiate between people with similarly pronounced characteristics.
The probability of solving an item changes depending on the characteristics of a person. The characteristics of the test subject are called personal parameters. Test subjects with low personal parameters (e.g. low IQ ) have a lower probability of solving a question than test subjects with high personal parameters, for whom the probability of a solution would be higher. That a person will solve a task is therefore never 100% certain, but also never with 0% improbable. The individual solution probabilities depending on the personal parameters are therefore mapped in a so-called item-characteristic function.
There are two types of adaptive testing:
- Branched Testing (from English to branch : to branch) With this test form, based on the successful or unsuccessful processing of a group of tasks, based on specified branching rules according to the manual, a decision is made whether the test person is to be given a more difficult or easier task group next (see Kubinger, 2009; page 107) (example: Adaptive Intelligence Diagnostic AID by Kubinger & Wurst, 1991)
- Tailored Testing (of English. To tailor : tailor): This test form is the theoretical skill level of volunteers, known person parameter, recalculated with each solved task, allowing the selection of "customized" tasks. Due to the computational effort, this form of adaptive testing can only be used in computer-based methods, e.g. B. the Frankfurt Adaptive Concentration Performance Test (FAKT) by Moosbrugger & Heiden, 1997.
See also
Individual evidence
- ↑ Amelang, M. & Schmidt-Atzert, L. (2006). Psychological diagnosis and intervention (p. 81). Springer: Heidelberg.