high jump adaptive testing 2

What is a Cutscore or Passing Point?

Nathan Thompson, PhDJuly 25, 2022

A cutscore or passing point (aka cut-off score and cutoff score as well) is a score on a test that is used to categorize examinees.  The most common example of this is pass/fail, which we are all familiar with from our school days.  For instance, a score of 70% and above will pass, while below 70% will fail.  However, many tests have more than one cutscore.  An example of this is the National Assessment of Educational Progress (NAEP) in the USA, which has 3 cutscores, creating 4 categories: Below Basic, Basic, Proficient, and Advanced.

The process of setting a cutscore is called a standard-setting study.  However, I dislike this term because the word “standard” is used to reflect other things in the assessment world.  In some cases, it is the definition of what is to be learned or covered (see Common Core State Standards) and in other cases it refers to the process of reducing construct-irrelevant variance by ensuring that all examinees are taking the testing in standardized conditions (standardized testing).  So I prefer cutscore or passing point.  And passing point is limited to the case of an exam with only one cutscore where the classifications are pass/fail, which is not always the case – not only are there many situations where there are more than one cutscore, but many two-category situations might use other decisions, like Hire/NotHire or a clinical diagnosis like Depressed/NotDepressed.

When establishing cutscores, it is important to use scaled scores to ensure consistency and fairness.  Scaling adjusts raw scores to a common metric, which helps to accurately reflect the intended performance standards across different test forms or administrations.  You may read about setting a cutscore on a test scored with item response theory in this blog post.  For a deeper understanding of how measurement variability can affect the interpretation of cutscores, be sure to check out our blog post on confidence intervals.

Types of cutscores

There are two types of cutscores, reflecting the two ways that a test score can be interpreted: norm-referenced and criterion-referenced.  The Hofstee method represents a compromise approach that incorporates aspects of both.

Criterion-referenced Cutscore

A cutscore of this type is referenced to the material of the exam, regardless of examinee performance.  In most cases, this is the sort of cutscore that you need to be legally defensible for high stakes exams.  Psychometricians have spent a lot of time inventing ways to do this, and scientifically studying them.

Names of some methods you might see for this type are: modified-Angoff, Nedelsky, and Bookmark.

Example

An example of this is a certification exam.  If the cutscore is 75%, you pass.  In some months or years, this might be most candidates, in other months it might be fewer.  The standard does not change.  In fact, the organizations that manage such exams go to great lengths to keep it stable over time, a process known as equating.

Norm-referenced Cutscore

A cutscore of this type is referenced to the examinees, regardless of their mastery of the material.

A name of this you might see is a quota.  Such as when a test is delivered to only accept the top 10% of applicants.

Example

An example of this was in my college Biology class.  It was a weeder class, to weed out the students who start college planning to be pre-med simply because they like the idea of being a doctor or are drawn to the potential salary.  So, the exams were intentionally made very hard, so that the average score might only be 50% correct.  They then awarded an A to anyone who had a z-score of 1.0 or greater, which is the top 15% of students – regardless of how well you actually scored on the exam.  You might get a score of 60% correct but be 95th percentile and get an A.

The following two tabs change content below.
Avatar for Nathan Thompson, PhD
My Twitter profileMy LinkedIn profileMy YouTube channel

Nathan Thompson, PhD

Nathan Thompson earned his PhD in Psychometrics from the University of Minnesota, with a focus on computerized adaptive testing. His undergraduate degree was from Luther College with a triple major of Mathematics, Psychology, and Latin. He is primarily interested in the use of AI and software automation to augment and replace the work done by psychometricians, which has provided extensive experience in software design and programming. Dr. Thompson has published over 100 journal articles and conference presentations, but his favorite remains https://scholarworks.umass.edu/pare/vol16/iss1/1/ .
/by