skip to main content
ACER

Understanding the science behind adaptive testing

Feature 6 minute read

Adaptive testing can provide a powerful solution for schools seeking accurate, meaningful insights into their students’ progress. In a recent webinar for international schools, Dr Jarrod Hingston, Director of School and Early Childhood Education Services at ACER, shared his expertise on how adaptive assessments work and how best to use them.

Adaptive testing provides personalised pathways for students

In conventional assessments, students answer the same, static questions on a given test. But in adaptive assessments, students’ responses determine the content (or questions) they see, resulting in personalised test pathways. 

‘It really is a method of assessment where the content is tailored to the ability of the individual,’ explained Dr Hingston. ‘Depending on the student’s responses, we can present new content that adjusts to their performance, eventually reaching a result that gives us a rich set of information about where that student is in their learning.’

This tailored approach allows teachers to gain more precise insights across a broad range of abilities, especially in classrooms with students performing at very different levels.

Adaptive testing supports each student to make progress

‘Our research shows that in most classrooms there can be 5 or 6 years of difference in ability levels,” said Dr. Hingston.

‘Adaptive assessment helps teachers understand exactly where each student is at, and what their next steps in learning should be.

For example, a high achieving student on a conventional (or ‘linear’/’fixed-level’) test, may end up with a report that looks like Figure 1. Educators can see some of what the student can do, but not what the next steps or challenges are for that student.

Figure 1: Individual report of a high-achieving student on a linear test, showing the student’s performance across each strand

In an adaptive test, a high-achieving student’s results might look more like Figure 2, showing clearly what the student has learned and where they need to be supported next.

Figure 2: Individual report of a high-achieving student on an adaptive test, showing the student’s performance across each strand

Adaptive tests can also be powerful for students on the lower end of the achievement scale.

‘We will find out more about what they can do,’ explained Dr Hingston. ‘This is also very important for the wellbeing of students, to be able to help inform them and give them some agency to understand what it is that they actually can do and to understand the point they have reached in their learning.’

When to use adaptive tests

While adaptive assessments are powerful, Dr Hingston was clear that they aren’t always the best tool for every purpose.

‘There’s an assumption that adaptive assessments are always the best approach, but it really comes back to the purpose of assessment,’ he said. For example, linear or fixed-level tests may be better suited for quick competency checks or entrance exams.

‘However, adaptive assessments offer valuable diagnostic information especially with a group of students of varying abilities.

‘In a single sitting, adaptive tests will give teachers a lot of information across a broad construct, say for example, in mathematics’ he explained.

The importance of quality

Finally, Dr Hingston encouraged educators to carefully consider the quality and design of any adaptive assessments they use.

‘The word "adaptive" doesn't necessarily equate to quality,’ he said. ‘It’s important to ask questions about the design, the purpose, and how the results are reported. The goal is to make evidence-based decisions that truly support student learning.’

Dr Hingston explained the importance of a well-formed construct and careful content calibration, including using reliable models to select test items.

It is also worth considering how the test uses its adaptive functionality to assess the whole construct and ensures content balance.

‘For example, we wouldn't want a student to sit an adaptive assessment where 80 or 90% of the items are assessing skills in the mathematical strand of Number and Algebra,’ Dr. Hingston said. This may happen with an item-by-item adaptive model.

‘And there's not really any assessment of algebra or patterns or chance and data. We want to get that whole picture of mathematics.’

A multi-stage model like the one illustrated in Figure 3 is an effective way of being able to assess achievement in a broad construct as students move through the testlets.

Figure 3: Three-stage testlet model of PAT Adaptive and PAIS Adaptive assessments

Learn more

You can watch the full webinar here. To find out more about adaptive assessments:

Subscribe to the Discover newsletter

Privacy policy