An interview with Mark Daniel, author of Growth Scale Value (GSV): Theory, Development, and Applications

April 6, 2023, 13:56 , 13:56 by

In testing and in research, you must have an effective and reliable way to measure change. While raw scores have typically been the go-to for this important role, Growth Scale Values (GSVs) are quickly becoming the preferred method.

Our Director of Pharma Sales, Lynsey Psimas, had the opportunity to speak with Mark Daniel, author of Growth Scale Value (GSV): Theory, Development, and Applications to discuss the importance of GSVs. He spent some time explaining GSVs and their importance in test scoring and pharma research.

Mark, tell us a little about yourself

My PhD program in experimental psychology focused on measurement, and even though content is always important in test development, I’ve put special focus on ways to structure, administer, and score tests to obtain as much useful information as possible in a reasonable time. Early in my career, I had the good fortune to work with Colin Elliott, who in the 1970s had developed one of the first clinical tests based on the Rasch model (a form of item response theory), the British Ability Scales. In creating the US version (Differential Ability Scales), we were able to make additional innovative applications of Rasch scaling, especially in adaptive and out-of-level testing.

Throughout my career, the Rasch model continued to be an important tool in developing diverse types of tests in cognition, neuropsychology, educational achievement, early childhood, speech/language, and behavior. Two decades ago, it was the basis for Growth Scale Values (similar to Woodcock’s W scores), which were originally designed to measure growth in school achievement, but which have also proven to be very useful in other change-oriented applications such as clinical trials. Currently I consult with Pharma researchers on behalf of Pearson to help them measure change most effectively, especially by using GSVs.

Can you explain GSVs in simple terms for those of us who are learning about them now?

GSVs can be thought of as an improved version of raw scores. Both GSVs and raw scores measure underlying ability; as ability increases, they both increase. However, the raw score increases faster or slower depending on how many items there are at a particular region of difficulty. If there are a lot of items, raw score increases quickly as ability increases, because there are many items on which the person can demonstrate their improvement. In a region with few items, raw score increases slowly. What this means is that a given increase in raw score corresponds to different amounts of increase in ability depending on the level of the scale where the change occurs.

GSVs attempt to remove this inconsistency. A GSV is an estimate of the ability level that would most likely have produced an observed raw score. Thus, GSVs adjust for the effects of the uneven distribution of item difficulties. By their nature, they attempt to be “equal interval”; that is, a given change in GSV reflects the same amount of change in ability at all levels.

Of course, GSVs do not perfectly achieve this goal. The Rasch model on which they are based is a simple one in which a person’s probability of success on a test item is based on just two things, the item’s difficulty, and the person’s ability. Nevertheless, they are a fundamental improvement over raw scores because they address the problem caused by the distribution of item difficulties.

In your opinion, why are GSVs so important to pharma customers and/or clinical trials?

The main reason is that GSVs are designed for measuring change because they attempt to be on an equal-interval scale. Not only is this helpful for interpreting an individual’s score change, but it also provides strong support for averaging change across people in the clinical trial who may vary widely in ability.

Two other features of GSVs are attractive to Pharma researchers. One is that each GSV has a score-specific standard error of measurement, which helps make statistical analysis of change more precise. The other is that a person’s GSV tells the probability that they will pass any particular item on the test. This means that a GSV change can be described in terms of the changes in what the person can do, which helps people interpret the meaning of a GSV change.

What are some of the benefits in using GSVs vs. something like normative scores?

GSVs and raw scores share an important characteristic that differentiates them from normative scores and age equivalents: they can be given to individuals with a wide range of chronological and developmental ages.

On most tests, normative scores can only be obtained for individuals whose chronological age is within the age range of the norms, and who perform within 2 or 3 standard deviations of the average for their age peers. Age equivalents are not restricted by chronological age, but they are available only for people with developmental ages within the age range of the test’s norms.

GSVs and raw scores can be given at any chronological age if the test’s difficulty is a good match to the person’s ability. For example, adolescents or adults with some neurodevelopmental disorders may have developmental ages suitable for a test normed on preschool-age children.

As long as the person’s raw score isn’t nearly perfect or nearly 0, the GSV or raw score can be useful for measuring change.

Not only does this feature enable a clinical trial to use the same test with individuals with widely varying ability levels, it also provides greater assurance that the same test can continue to be used with each person even if their score changes substantially over the course of the trial.

How do GSVs compare to other outcome measures?

GSVs are tailored for situations where measuring change is the main concern. Clinical trials are an unusual application for tests because they involve repeated administrations over months or years, something that is rarely done in clinical work (other than in education). So, while GSVs are optimized for this purpose, they give up some other applications. For example, they don’t give normative information, so if a study needs scores that can be used for diagnosis or inclusion criteria, GSVs are not sufficient.

GSVs are like raw scores and age equivalents in that they all reflect absolute performance— what the person actually does. This characteristic makes change simpler to interpret. If an absolute score increases, it means that the person performed better on the tasks (test items).

In contrast, change in age-based normative scores such as standard scores and percentiles is more complex to interpret because it means that there was change in how the person’s performance compares to the average performance of their age peers. A normative score can decline even if the person’s absolute performance improved if the improvement was slower than average.

GSVs differ from other score types by offering equal-interval scaling, score-specific SEMs, and probabilistic interpretation.

Thanks so much Mark for taking the time to answer our questions as well as provide knowledge, insight, and clarity around the importance of GSVs.

To learn more about the use of GSVs in research and testing, download Mark's Growth Scale Value (GSV): Theory, Development, and Applications report.

Mark Daniel, PhD

Mark’s PhD program in Experimental Psychology (Texas Christian University, 1984) emphasized quantitative methods and measurement. For 36 years he developed psychological and educational tests for The Psychological Corporation, AGS, and Pearson in the roles of Research Director, Director, and Principal Researcher. He is a Fellow of the American Psychological Association (Division 5). Mark now consults with Pearson Pharma Research Services on ways to enhance the utility of clinical tests for treatment studies, with a focus on applying new methods of scoring and interpretation.

Lynsey Psimas, PhD

Dr. Psimas is a licensed Clinical Psychologist with extensive experience in the assessment of children, adolescents, and adults. She has worked in private practice, therapeutic schools, and as a professor. She specializes in neuropsychological assessment, autism, ADHD, and learning disabilities. Dr. Psimas lives in Houston, Texas with her family.

Pharma Research Services

Supporting clinical research with an 80 year history in measurement science

Learn more about Pharma

View Pharma blogs and research

Professional assessments

Digital solutions

Other assessment types

Featured topics

Training

Research

Webinars & Blogs

Ordering

Account information

Manage orders

Contact us

An interview with Mark Daniel, author of Growth Scale Value (GSV): Theory, Development, and Applications

Pharma Research Services

Supporting clinical research with an 80 year history in measurement science

SHOP ASSESSMENTS

LEGAL POLICIES

HELP & SUPPORT

ABOUT PEARSON