Overview: An adaptive test uses a computer algorithm to adjust question difficulty according to an individual’s performance on a particular test. When you do well, you’ll get tougher items. When you do poorly, you get easier items. Moreover, in adaptive tests, students’ performance is measured in the moment so that a score can be determined quickly based on their performance.
While this may seem simple and logical at first glance, there are several technical challenges to consider. Over the past half century, researchers have developed several approaches to how an adaptive algorithm can work to address these challenges.
- Adapt the difficulty for each item separately.
- Adapt the difficulty to each block of items (sections), also known as multi-stage testing.
- The test can be adapted entirely differently (for example, using random forest or cognitive diagnostics).
Graduate students have been taking adaptive tests for a long time. Adaptive tests, such as the Graduate Management Aptitude Test (GMAT), change difficulty level based on what you answer each time. There are other tests, such as the digital SAT, which measure a test-taker’s performance across multiple questions.
Online SATs are the first adaptive tests high school students will take, so they will have to adjust their usual testing strategies.