What is Adaptive Testing?
Adaptive Testing, also known as Computerized Adaptive Testing or Tailored Testing, is a method of using algorithms and machine learning to make tests more optimal, effective, and efficient.
Every time a candidate answers a question, the test analyses the response and based on it, decides the difficulty level of the next question. The test, with each response, also determines the number of questions required to assess the mastery level of the candidate.
At the beginning of the test, a candidate will start with a medium difficulty question. If he/she answers it correctly, the next question will be more difficult, or easier if answered incorrectly. Adaptive tests do not have a fixed number of questions; they end when the termination criterion is met, which means that as soon as the test has determined the ability of the test taker and has marked him/her accordingly, it ends.
Advantages of Adaptive Testing
- Accurate scoring: Data collected for standard paper-pencil tests over the years shows that they provide the best precision for only medium-ability test takers. Test takers with extreme abilities are the most affected. In contrast, adaptive testing is fairer because the test designer has customized control over score precision for each test taker. Hence, it is said to be more precise from a psychometric point of view.
- Increased psychometric efficiency: Similar to pedagogy, it is tough to engage and analyse high ability examinees with extremely trivial questions and lower ability examinees with difficult questions. Adaptive testing doesn’t administer such questions, significantly reducing the number of questions required to test a candidate. As a result, the amount of time needed to test each candidate optimally is reduced substantially. The seat time savings cost can be between 50 to 90 per cent. In addition to time being saved, it also increases candidate engagement. A higher ability candidate will not have to waste any time solving easy questions and can concentrate all his mental energy on the questions specially designed for him/her without ever feeling bored.
- Security: Since each candidate receives a unique set of questions, there is very little chance of cheating. Even in the case of re-examination, candidates are given a different set of questions from the question bank.
- Cost: Adaptive tests use a technology called Item Response Theory (IRT) for the design process. IRT is a way to analyze responses to questions to improve both accuracy and reliability. While investigating a candidate’s responses, IRT looks at not only the number of correct answers the candidate has answered but also the level of difficulty of those questions. The SAT and GRE both use IRT for their tests.
IRT is used to convert adaptive tests from fixed-length tests to variable lengths. This means tests can be finished in lesser time, with lesser number of questions, and with higher accuracy. Test time can be cut down to as much as half, leading to a substantial decrease in costs.
Although the engagement and uniqueness of questions provided to test takers are high, it isn’t straightforward for them to complete a timed test. If a person with high ability ends up spending too much time on one difficult question, he/she will not be able to keep aside enough time for the next set of questions and will end up handing over an incomplete answer sheet. An untimed adaptive test is also a tough option for high-stakes summative assessments for aptitude jobs.
In conclusion, adaptive testing is beneficial for both students and instructors. Students get a chance to personalize their learning experience, be responsible for setting their own pace and identify their problem areas. As for the instructors, it is easier for them to work in mixed-ability large classrooms, make better instruction plans for students, and make optimum use of the classroom time and resources.
Explore more at: TCS iON