Build a clear scoring rubric

Choose criteria before point values

Start with the learning goals the assessment is intended to demonstrate. Turn them into a small set of distinct criteria. For an essay, those might include ideas and evidence, organization, and expression. Avoid scoring the same feature in several places unless that overlap is intentional.

The supplied templates are original examples to edit. They are not validated institutional rubrics. Replace their language with descriptions appropriate to the assessment, course and learners.

Describe evidence at each level

Write what a marker can observe at each achievement level. “A clear position supported by relevant, explained evidence” is more actionable than a label such as “excellent” on its own. Labels are useful shorthand, but descriptors should do the explanatory work.

Assign points to each level. In the builder, selecting a level enters its point value; you can also type a score manually. The tool checks that level points do not exceed the criterion’s maximum.

Choose points or explicit weights

In points mode, a criterion worth eight points has twice the influence of one worth four. To give criteria a different fixed share, choose weighted mode and make the weights total 100%.

With scores of 3/4 and 2/6, total-points scoring produces 5/10 = 50%. Equal 50% weights instead give 75% × 0.5 + 33.333…% × 0.5 = 54.166…%. Make the chosen method visible before the rubric is used.

Try the rubric, then reuse it

Apply the rubric to a few example responses and check whether the descriptors distinguish the evidence you intended to assess. If several markers use it, compare how they interpret the same examples and clarify ambiguous wording.

Save the definition on this device, print a blank copy for marking or sharing, and clear the scores for the next assessment. Use a JSON backup to preserve editable levels and descriptions outside the browser. CSV provides the score summary; it is not a replacement for the editable backup.

• Share criteria and scoring expectations before the assessment.
• Keep descriptions specific and understandable.
• Avoid student names and identifiers in local workspaces.
• Check the scale and rounding settings before recording a result.