Using percentiles is a ubiquitous method of evaluating and communicating students’ academic performance throughout their primary, secondary, and tertiary education. However, despite their prevalence, many scholars and educators such as Thomas Guskey eschew the utility and accuracy of assigning student performance to a 100-point scale, instead calling for a return to the arguably more informative and manageable band-scores.
There is more than a century of literature critical of using percentiles to illustrate and assign student performance. In 1912, Wisconsin researchers Daniel Starch and Edward Elliott had 147 high school English teachers grade two identical papers; more than 30 different percentage grades were assigned, ranging from 50 to 97. Replicating this study the following year with 128 math teachers, Starch and Elliott found even greater score variation with grades ranging from 28 to 95 percent.
While not assigning causation to Starch and Elliott’s studies, significant pedagogical and academic discourses sparked by these types of studies during the 20th century, which saw a gradual shift from percentage grades to band scores such as the three-point (Excellent, Average, and Poor) and five-point (A, B, C, D, F) scales.
Percentage grades have made a comeback as the preferred grading scale, due primarily as Guskey argues, to the growth of digital tools and learning management systems built to support the scale preferred by the engineers and technicians who designed them: percentiles.
There are several logistical, reliable, and motivational concerns surrounding percentage grades. As referenced above in the Elliott and Starch studies (1912, 1913), intergrader reliability and grading consistency may not be as reliable when assigning students to a single point on a scale of 100. Furthermore, identifying half or more of the points as failing may instill a self-fulfilling prophecy or learned helpless in students; going from a B to an A requires an increase of 10 points, however zero to 50, 60, or 70 may seem insurmountable.
The path forward, Guskey suggests, is a return to band scores which were the prevailing practice of assigning and reporting grades throughout the majority of the 20th century. Limiting the number of grade categories to three or five may increase reliability and consistency while providing students and stakeholders with more meaningful and actionable information on their performance.
Read more:
About the Author: Dustin is a senior account manager with DesignedUX, providing communications and marketing strategy to organizations in education and technology. Dustin is also a part-time faculty member at Centennial College and serves on the board of the Canadian Public Relations Society