What is the least enjoyable part about being a professor? For me, the answer is easily “grading.” For years I dreaded the whole process – determining whether a response was worth 4 points or 5, ensuring consistency across students, and arguing over partial credit instead of discussing course content. Opposite this dread was the knowledge that one of the most important roles we have as educators is providing feedback to students. Since reading Nilson’s (2015) book on Specifications-Grading, I have implemented some variation of the system in all my courses. While I don’t suddenly love grading, I have been convinced this is a better approach; I spend less time grading; and the quality of the student work has improved! In this post, we’ll discuss the key components of Specifications-Grading; in a follow-up post, we’ll discuss an implementation for an intermediate statistics course.
While there is no one way to implement the Specifications-Grading framework, there are three defining characteristics:
- A student’s course grade is earned through the successful completion of a pre-determined collection of assessments.
- Each assessment is graded pass/fail.
- The requirements for successfully completing an assessment are clearly communicated to students in advance.
Setting up the Hurdles
Traditionally, students earn grades defined by a weighted average of points. In contrast, Specifications-Grading promotes linking the grades directly to student learning objectives.
While course design is an extensive topic on its own (see Wiggins and McTighe  for an overview), we might summarize it as: the plan stating what skills the students develop in the course, how students refine those skills, and how those skills will be assessed. Those skills that are developed are known as course learning objectives; course activities and assessments are then tied to these objectives.
For example, a course-level objective might be:
Given a statistical analysis, interpret the results in the context of the study objective.
This broad skill is supported by several module-level objectives such as:
Given the least squares fit of a simple linear regression model, interpret the estimate of the slope and intercept in the context of the problem.
This skill is practiced on an in-class worksheet and assessed using a short-answer question on an exam. There are various frameworks for learning objectives (see Wiggins  and Fink  for two examples); regardless of the framework you choose, selecting verbs that accurately describe the skill we intend to assess is critical.
Nilson outlines three possibilities for linking objectives to course grades:
- More Hurdles: Higher grades reflect mastery of more objectives.
- Higher Hurdles: Higher grades reflect mastery of more advanced learning (think further up Bloom’s taxonomy).
- More and Higher Hurdles: Higher grades reflect mastery of both meeting more objectives and more advanced objectives.
Each grade is defined by a “bundle” of assessments which must be completed. When making this link, it helps to ask the following questions:
- What should every student who leaves my course be required to have mastered? This defines the requirements for a D.
- What material distinguishes proficiency from expertise? This might define the difference between an A and a B in the course.
Note that a student seeking a C in the course may not attempt all assessments in the course; that can be hard for faculty. The trade-off is we only grade work for students who want to complete an assessment, and we have a much better idea of what each grade in the course represents.
While I have implemented some hybrid schemes, a true Specifications-Grading system grades every assessment pass/fail. From a faculty perspective, pass/fail grading can immensely reduce the time spent grading. Instead of debating the points on a problem, we only mark whether the work is satisfactory or unsatisfactory. From a student perspective, this requires a shift in mindset from throwing the kitchen sink at a problem and hoping for partial credit to producing high-quality work. The grade the student is aiming for in the course determines which assessments must be completed at this high level.
For as many fears of pass/fail grading as we may have, students will have more. To ease these fears, I use the phrases “successfully complete”/“do not successfully complete” instead of pass/fail when describing the framework to students. I also emphasize that “successfully complete” does not mean perfect, which brings us to the heart of Specifications-Grading – defining “successful completion.”
Defining “Successful” Completion
When we mark something “successfully completed,” we are really assessing whether the student’s work meets the corresponding specifications, one or more pre-determined requirements that we set for that assessment. The pass/fail aspect means there is no partial credit for meeting only some of the specifications.
The specifications can be very detailed or broad; the requirement is they define what we consider to be minimally acceptable work to meet the learning objectives being assessed. As an example, the specifications for my homework assignments in Introductory Statistics are:
Every problem must be attempted and the solution, even if incorrect, attempts to address the problem using course material.
Notice students do not need to get a single question on the homework correct in order to receive credit for the assignment; they simply need to complete the assignment. This reflects my course design, where homework is practice meant to inspire discussion of course concepts. Real assessment happens on weekly quizzes, where the specifications set a much higher bar to demonstrate mastery. For example, a quiz question asking students whether a study provides evidence against the null hypothesis given a graph of the null distribution could have the following specifications:
- State whether you believe there is evidence or not.
- Justify your position using the provided graphic.
- Use consistent and appropriate logic when applying the graphic.
- Do not use statistical terminology discussed in class incorrectly.
While a similar question was on the homework, the quiz question is assessed more rigorously. When grading, assessment is limited to only those things described in the specifications. The most difficult part of constructing specifications is restraining ourselves; in writing specifications for a report, for example, we should not hold students to the same criteria we would hold our peers.
Specifications-Grading only works if the specifications are clearly communicated to students in advance. This is where a bulk of preparation goes. I strongly recommend giving examples of both acceptable and unacceptable work and annotating these examples to specifically describe how the work met or failed to meet the specifications.
Adjusting to the System
Life happens, and students’ priorities may not always align with the class. Instead of compromising your expectations for student work, you can incorporate flexibility while making the assessment itself a learning experience for students. One way of doing this is to provide multiple opportunities to demonstrate mastery (such as giving 3 quizzes on the same topic, but only requiring students pass 1); the other is through “tokens” (I prefer Dr. Megan Heyman’s “outliers”). Students are given 2 opportunities, for example, to declare an assessment an “outlier”; these assessments are then eligible to be revised and resubmitted. Or, a student might declare an assessment an outlier to receive an extension with no questions asked. Such flexibility allows students to improve upon unsuccessful attempts.
Remember, you define what students must complete to earn each grade and what constitutes “satisfactory” work. Once the class begins, your role is helping students overcome these hurdles. In the next post, we’ll see an example.
Contributing author Eric Reyes is an Associate Professor in the Department of Mathematics at Rose-Hulman Institute of Technology and has been tinkering with specifications grading in his statistics courses for the past five years.
Fink, L. D. (2003). Creating significant learning experiences. Jossey-Bass.
Krathwohl, D. R. (2002). A revision of Bloom’s taxonomy: An overview. Theory Into Practice, 41(4), 212–218.
Nilson, L. B. (2015). Specifications-Grading: Restoring Rigor, Motivating Students, and Saving Faculty Time. Stylus.Wiggins, G. & McTighe, J. (2005). Understanding by design (2nd ed). ASCD.