November 2010

There are many ways to report test scores. It is important to use an approach that is accurate and understandable to students, teachers, administrators and parents.

The most common way test scores are reported is as raw scores. Raw scores are simply a tally of the number of questions answered correctly.

Raw scores are easy for students and others to understand. And, we are very accustomed to receiving results that tell us how many questions "we got right."

But raw score reporting may not be the best choice. Even though raw scores are familiar, they may not be the most accurate way to report scores; reporting scores on a scale may be more accurate.
A detailed discussion of scaling and measurement is well beyond this article, but there are a couple of key points that can help clarify why it is often better to report scaled scores.

Raw scores are limited in meaning. Any test is a "proxy" for finding out where a student stands on a "true" underlying scale reflecting the domain you are measuring. So, if you are administering a 20 question reading test, that test is only a proxy for reading ability; the raw scores from that reading test are only a rough indicator of a person's underlying reading ability. By using advanced measurement techniques, it is possible to change the raw scores into an underlying scale that better reflects the underlying reading ability.

Raw scores do not take item difficulty into account. Since the questions on each test are not equally hard and may differ statistically from each other in several other ways, the distance or "amount of ability" between scores on the raw scale may be greater at certain points. For example in terms of actual reading ability, the distance or "amount of ability" between 10 and 15 correct may be greater than the distance or the "amount of ability" between 15 and 20 correct on the test. Converting to scaled scores can help eliminate these differences, by making sure that the amount of ability between each point across the whole scale is the same and one score point difference in the middle means the same thing as one score point difference near the top of the range.

Raw scores should not be compared over time or across forms. Reporting scores on a common underlying scale (rather than raw scores) is particularly valuable when your testing plans call for the use of more than one form of the test. This includes using multiple forms at a single administration or using different forms of the test to monitor growth or progress over time.

Using raw scores under these circumstances can be very misleading. Despite your best efforts to make two forms similar, it is likely that they vary in difficulty. Take our reading test for example; Form B may be harder than Form A. Let's say this is a 20 question test and that a raw score of 10 on Form A may be equal in difficulty to a raw score of 8 on Form B. A student who takes Form A and gets a 10, then takes Form B later in the year and gets a 10 again, appears to have no growth. But in fact, because of the difficulty of the two tests there was improvement in performance. If both tests had been reported on the same scale, this growth in the scores would be readily apparent.

A detailed discussion of scaling is beyond the scope of this article. But, you should be aware that there are many different types of scales available--all with their own strengths and weaknesses.

If you currently provide assessments or are developing new ones, please contact us to discuss what type of reporting makes the most sense for your program.