Adaptive ability test faster and more accurate
Many of NOA's customers are familiar with the Multicultural Capacity Test, a version of which exists for secondary and higher education levels. A proven method that will remain available in the future. But in the meantime, there is already the next step in the development in the form of an adaptive capacity test. A test that responds to the level of the candidate during administration and thus produces "tailor-made" results. This is faster and more accurate.
Michiel Hol, researcher and developer at NOA, has been working on the development of this adaptive test in recent years. "There is nothing wrong with the capacity tests we already have, the MCT-H for higher education level and the MCT-M for intermediate education level, but we were still looking for a test that better suits the individual candidate so that he becomes faster and more precise, while also remaining suitable for larger tests. That has led to this new adaptive MCT-A test that I expect our customers will be happy with." The test will soon be available on the NOA Talent Portal, NOA's new platform.
How does such an adaptive test work? The principle is that the candidates are asked questions based on his or her previous answers. For example, if someone is good at math and fills in the initial questions correctly and quickly, the test program registers that, after which it chooses follow-up questions that are more difficult. Until they become too difficult for the candidate and a step back is taken until the right level is reached. Conversely, the program also registers it when the initial questions are too difficult, then the level is adjusted downwards. Michiel: "In this way, the level of the candidate in question can be determined more quickly."
"An adaptive test is not AI, it is based on an ordinary statistical model," Michiel emphasizes. For the adaptive capacity test, all calculations are first prepared in the statistical program R. With adaptive testing, all kinds of calculations are performed that have been worked out in advance during the completion of the test. The programmers then built all calculations into the NOA Talent Portal." This was followed by the first real testing. But that's a lot of fun to do because then you see if it works as you hoped. And in our case: it worked! So we had a big Hurray! moment."
The next step was to analyse the existing questionnaires and develop new questions. For the new adaptive test, the questions from the existing capacity tests for higher and secondary education level could be combined because the test adapts itself to the level of the candidate. Differentiation is therefore no longer necessary. "The operation of all those existing questions had to be investigated," says Michiel, "because they are administered in a different way in the new adaptive test. And we also added new questions. As a result, the MCT-A can now choose from a larger number of questions than the existing MCT-H and MCT-M contain. This has the additional advantage that the questions will be known less quickly. Moreover, cheating on this test will be very difficult because the candidates for the same subtests will often be asked different questions."
To investigate the effect of the new questionnaires, a total of 12,000 tests were taken. Michiel: "Recruiting the candidates for this was not easy because the samples had to meet many requirements to be representative. Consider, for example, the distribution in terms of age and education. This part took a long time, was difficult and time-consuming. Fortunately, we had the support of Qualtrix in this."
"I am now in the final phase before the MCT-A is operational, namely the analysis of the studies and the determination of all institutions. In it, you choose which items will be eliminated and how long the adaptive subtests will be. In the meantime, new instructions are being worked on and new reports will also be made. And eventually test again. What always remains exciting: what does the result look like?"
|
Advantages of the MCT-A:
|
