Download the blank rubric and original example
Editable tab-separated files for a blank three-level rubric and a separate constructed literature example. They contain no student records or grading decisions.
Use a blank record for your own work. Keep examples separate from your evidence and follow the instructions below.

On this page
Interactive rubric builder
Draft an analytic rubric
Set the task and learning objective, then describe what evidence looks like at each level. Edit the original interpretation example or start with your own criteria. Your changes stay in this browser tab; do not enter identifiable learner information. The wording is a draft for your course, not a validated assessment instrument.
What an analytic rubric makes visible
An analytic rubric separates an assignment into criteria and describes performance for each criterion across levels. This makes it possible to distinguish, for example, accurate use of evidence from the quality of reasoning about that evidence. A checklist may record whether a component is present; an analytic rubric also describes differences in the quality of that work.
The builder starts with a task and learning objective, then lets you edit level names, criteria and descriptors. It does not select learning goals, assign grades, recommend weights or determine whether the rubric matches your department’s requirements. Those decisions stay with the instructor.
Write descriptors that distinguish performance
Begin with the evidence a learner can produce, not a trait such as “excellent,” “creative” or “clear.” Name what a reader would see: a relevant quotation, an explained connection, a correct method, or a stated limitation. A stronger descriptor tells students how the work differs at this level; changing an adjective while leaving the same evidence in every column does not create a useful distinction.
Cornell’s rubric guidance recommends criteria that can be assessed, descriptors that distinguish levels and feedback on a draft before use. UIC’s teaching guide also describes an analytic matrix as separate criteria crossed with performance levels. These principles inform the builder, but this site does not reproduce or endorse either institution’s rubric.
A worked example: interpret a character’s decision
Scroll the table sideways to view all columns.
| Passage detail | What it supports | What remains unknown |
|---|---|---|
| Mina folds the letter and places it beside a sealed envelope. | The letter and envelope are physically near one another. | The passage does not say what either contains or who sent them. |
| At the doorway, Mina turns back and moves the lamp closer to the page. | She turns back and moves the lamp closer to the page; wanting to read is one plausible inference. | The passage does not state why she moved the lamp or whether she read the page. |
The original example in the builder uses a short constructed passage: Mina folds a letter, leaves it beside a sealed envelope, then returns to move a lamp closer to the page. Students compare two plausible readings of that action, support one with details and name what the passage does not establish. The example is invented for this exercise, not a published literary text or a student response.
The three criteria separate source fidelity, interpretive reasoning, and uncertainty and scope. Their descriptors describe observable evidence rather than a single required interpretation. Before using them, adapt the task to the text and course objective, then check whether two readers can apply the distinctions consistently.
Compare three constructed responses to see where the reasoning changes. A stronger response says: “She may have returned to read the page: moving the lamp closer would help her see it. She might instead have checked whether the envelope was secure, but the passage never says she touched it. Reading fits the final action better, though the text does not say what the page contains.” It weighs direct evidence against an alternative and keeps the conclusion tentative. A partial response says: “She might have returned to read the page or check the envelope. She moved the lamp closer to the page.” It names both explanations and one detail but does not compare how well the detail supports either. A weak response says: “She went back because the envelope held a secret letter from her sister.” The passage gives neither that relationship nor that content. These examples are written for calibration, not taken from student work or a grading trial.
Calibrate the draft before grading
Test the draft against two different examples of work, ideally with another instructor or teaching assistant. Ask which descriptor fits each example and point to the evidence that supports that choice. If readers choose different levels for the same reason, clarify the criterion or describe the boundary between levels. Then revise and repeat the check.
This is a review step, not proof that a rubric is valid or reliable. The builder cannot judge a student’s work or establish that different graders will agree. Keep the instructions, rubric and feedback process aligned with the learning objective, and explain the criteria to learners before they submit.
Questions about this topic
What is an analytic rubric?
An analytic rubric separates an assessment into criteria and describes performance at multiple levels for each criterion. It makes dimensions of a task visible individually; a holistic rubric gives one overall judgment instead.
How many criteria should a rubric have?
Use the number that makes the important learning goals assessable without fragmenting the task into minor checks. There is no universal count. Combine overlapping criteria and remove items that do not align with the assignment.
Does this rubric builder assign grades or points?
No. It creates editable criteria and performance descriptors. It does not add weights, calculate scores, grade submissions or decide what a course requires.
Can I use the example rubric as written?
Treat it as a constructed model of descriptor differences, not a ready-made assessment. Change the task, evidence, criteria and wording for your course, then review it with colleagues and sample work before sharing.
Is the rubric validated?
No. The builder checks for blank or repeated wording but cannot establish validity, fairness, reliability or alignment. Instructor review and course-specific calibration are still needed.
Sources and editorial notes
The supporting reading below informs our editorial guidance. Exercises and scenarios illustrate a method; they do not report a completed product test or measured learning outcome. Read our editorial policy or send a correction with a supporting source to info@ai-essay.com.