Rubric maker

How to use the rubric maker

1. Pick a starting point: a coding project, a presentation, a written report, a group project or a blank grid.
2. Name the criteria (rows) and the levels (columns), and give each level its points. Weight a criterion ×2 or ×3 if it should count twice or three times.
3. Write a descriptor in every cell: what the work looks like at that level.
4. Read the checks. They flag empty cells, levels that say the same thing, and descriptors that judge without describing.
5. Print it with a score column, copy it as Markdown, or download it as a CSV file for a spreadsheet.

Analytic or holistic?

This tool makes analytic rubrics: each criterion gets its own score, and the scores add up. A holistic rubric gives the work one overall level, with one description per level.

For training, analytic usually wins: the point of the mark is what the learner does next, and that needs one score per skill. Start your criteria from what the course promised. If you’ve written objectives with the learning objectives generator, each criterion should match one of them.

Write descriptors that describe the work

The most common weakness in rubrics is a descriptor that only judges: “Good use of tests”, “Excellent presentation”, “Adequate referencing”. Two markers will read “good” differently, and a learner can’t tell what to do to move up. Susan Brookhart’s guide to rubrics (2013) puts it simply: describe the performance, don’t just evaluate it. Write what you would see in the work.

• Write the top and bottom levels first, then fill the middle. It’s easier to describe the steps once you know both ends.
• Make each cell readable on its own. “Better than Developing” only works if the reader has the whole row in front of them.
• Change one or two things per step. If every level differs in five ways, no one can place work that is strong in some and weak in others.

How many levels?

Three to five is the usual range, and that’s a rule of thumb rather than a finding. Four is a common choice: there is no middle level to settle on when unsure, and few enough levels that the difference between neighbours is visible in the work. Every extra level is one more line two markers have to agree on. Three works well for quick, formative checks; go beyond five only if you can describe each step clearly.

Share the rubric before the task

A rubric that learners first see next to their mark is a marking aid. Shared with the task, it becomes a teaching aid. Two reviews of the research (Jonsson & Svingby 2007; Panadero & Jonsson 2013) found that rubrics help learning mainly by making expectations clear, and that the benefit comes when learners use them while they work.

• Hand it out with the brief, and walk through one row together.
• Ask learners to place their own draft on it before they hand in.
• Use it for peer feedback: one criterion per reviewer keeps comments specific.
• Mark with the same sheet, so the feedback points to a cell they’ve already read.

A worked example

The coding project template has four criteria and four levels worth 4, 3, 2 and 1 points. Functionality counts twice, so the total possible is 4 × 2 + 4 + 4 + 4 = 20 points. Here is one learner’s project:

The score says where the project stands; the rubric says what to do next. Testing is at Developing (“a few tests, or some of them fail”), and the next cell along is the feedback: “Tests cover the main features, and they all pass.” If you turn scores into grades, set the boundaries before you start marking. For quick checks during the course rather than at the end, a single well-written question does more than a rubric: try the multiple-choice question checker.