Criteria define the distinct elements of expert or competent performance of the tasks central to the assignment. As the number of criteria increases, so does the amount of time required to review assignments. Too few criteria can lead to a rubric that does not effectively assess the range of knowledge, skills, and attitudes required to complete the task assigned. Generally, 4 to 6 criteria assess the breadth of competencies that are most essential to an assignment. 

A single criterion can be used to create a holistic rubric with very general descriptions. Holistic rubrics do not provide targeted feedback and research suggests they are less consistently used. Holistic rubrics work well to speed grading for low stakes activities that only need very general feedback (e.g., discussion forum posts or responses). For more significant assignments, an analytic rubric with multiple criteria is more useful for reviewers and students. Effectively selecting the most important criteria is the first step to designing effective analytic rubrics.

Once the most essential criteria for competently completing the assignment have been identified, different quality levels need to be identified. Selecting the number of quality levels is a critical decision. While a greater number of quality levels allows for finer distinctions, more levels increase the time required to develop the rubric and to review assignments. In addition, research has shown that as the number of quality levels increases, consistency across graders or reviewers decreases. Considering the ultimate grade distinctions that will be made in the course can help determine the appropriate levels of quality for significant assignments.

The labeling of quality levels requires careful reflection. In learning contexts, instructors typically distinguish levels of competence, mastery, or expertise. This framing emphasizes a developmental teaching perspective and communicates a growth mindset. Instructors should beware of quality descriptors that demoralize students (e.g., incompetent, barely adequate, almost competent).

The most equitable rubrics create a detailed table describing the key features for each criteria at each quality level. Criteria are listed along the left-most column (often according to hierarchy of importance or process order) and quality levels are arranged across the top row of the table (either from low to high or high to low). Each of the remaining table cells is filled with a description of key features that can be observed for the specific criteria at that quality level. These often focus on key factors that represent bottlenecks in student learning or critical steps to increased levels of competence.

Rubrics are flexible tools and instructors use a range of strategies to score student work using rubrics including:

1. Setting weights for each criterion, and single scores for each quality level. This approach speeds grading and minimizes discretion that might be a source of bias.  Many digital tools support this strategy.
2. Weighting criteria and providing a range of scores for each quality level. This approach supports instructors interested in making more fine-grained distinctions.
3. Focusing on the overall combination of quality levels across criteria to assign a grade. This is a simplified grading structure that focuses on the overall grade and holistic judgment of the instructor or grader. For larger course enrollments, this strategy increases the risk of inconsistent or biased grading.
4. For drafts and formative assessments, focusing on providing students feedback with the rubric and simply assigning a complete or incomplete grade can be an efficient and effective strategy

Consider what, exactly, you want students to learn from the assignment. Write this down; it will guide the creation of criteria for your rubric.

Decide how you want to grade students. What elements of the assignment do you want to give them feedback on? This means determining the criteria associated with the task. To determine the criteria, think again about the goals of the assignment. What do you want students to accomplish? What do you want them to learn? Keep these criteria descriptions brief. Also, try to have an even, rather than an odd, number of criteria. This prevents the middle criterion from becoming a catch-all, allowing for more nuance in grading.

Decide whether you will use a letter grade, percentages, points, a rating scale, or some other scoring method in your rubric. How will you label them? With numbers or descriptive labels? 

Make sure the descriptors follow a logical progression. That is, descriptors indicating poor performance should be distinctly different from descriptors indicating high performance. And there should be consistency within the descriptors, meaning they should focus on particular attributes that carry through all criteria. 

Rubrics offer a more objective means of assessing student work, but that doesn’t mean they should assume a negative tone or offer an overly pointed critique of the learner. Try to refer to the assignment rather than the student when developing criteria. Avoid overly subjective language and use active voice where appropriate.

Test your rubrics with a variety of scores for each criteria and see how significantly an outlier in one criterion will impact the overall grade. Consider whether adjustments to the weighting of criteria or the way points are allocated would more accurately reflect the appropriate grade.

Asking a colleague to review the rubric in advance is one of the best ways to ensure that your expectations are clear. You can ask colleagues to focus on specific elements of the rubric or to provide overall feedback.