How do you ensure that an assessment result is truly a fair reflection of what a candidate is capable of? The answer lies not only in the quality of the questions, but also in the cut score. Yet in practice, cut scores are still surprisingly often chosen arbitrarily or simply set at a standard threshold such as 55 percent.
Monika Vaheoja, data scientist and psychometrician, regularly writes in her blogs about fair decision-making in assessment. In her blog Angoff Method: How Do You Determine a Fair Cut Score?, she makes a sharp observation: “A fixed threshold of 55% is not a cut score policy. It is postponing a substantive decision.” In doing so, she highlights the importance of setting cut scores based on content, learning objectives, and performance expectations.
Why a standard percentage does not always work
Beyond determining a well-founded percentage, there are many other important choices involved in setting a cut score. A single cut score, for example X% of the total available points, may seem straightforward, but it often fails to reflect the complexity of an assessment. Not every component is equally important, and not every part of an assessment should be compensable by performance on another section. When everything is merged into one total score, there is a risk that these differences become flattened.
With a single cut score, all points are added together and the final total determines whether a candidate passes. In many cases this works perfectly well, but there are also situations in which this approach falls short. Consider assessments where certain components are far more critical than others, or situations in which weak performance on essential competencies can easily be compensated for by strong performance elsewhere. In addition, a total score alone may provide insufficient insight into a candidate’s underlying knowledge and skills.
A concrete example is a candidate who scores poorly on application-based questions, but still passes because of high scores on reproduction-based questions (often multiple-choice or recall questions). Formally, the result may be correct, but the question remains whether this outcome is actually desirable. To quote Monika once more: “A content-anchored cut score answers a different question: what can the student who scores just enough actually do? And is that enough?”
Perhaps it is acceptable for a candidate to pass based primarily on reproduction questions. But what if that was never the intention? What if the candidate was also expected to demonstrate the ability to apply knowledge in practice? In that case, the cut score should reflect this requirement. For example, by assigning greater weight to application-based questions, or by requiring a minimum score on those questions in order to pass the assessment as a whole.
From a single threshold to layered assessment
This is where a composite, or layered, cut score comes into play. In this approach, an assessment is divided into separate components, each with its own cut score. In addition to defining cut scores per component, different weights can be assigned to reduce the possibility of compensation between sections.
If additional weighting is still insufficient because knowledge in a certain domain is considered critical, a passing score can be made mandatory for that specific component. This means that without passing that section, a candidate cannot pass the overall assessment. This may be particularly relevant in sectors where competence in specific domains is essential, such as healthcare, aviation, and public safety.
Besides these assessment-related advantages, layered cut scores also benefit students. When they are allowed to review their results, they can see how they performed on the underlying components and identify where additional attention or improvement is needed.
A cut score that supports fair outcomes
A layered or composite cut score helps assessment coordinators create fairer and more transparent assessments. It prevents undesirable compensation effects and ensures stronger alignment with educational objectives. In short, it is a powerful way to improve assessment quality.
.webp)