If you've ever taught a multi-section course, you know the conversation. Two of you grade the same assignment, compare notes, and discover you're a full letter grade apart on a paper you both read carefully. You reread it together, trying to reconcile. You can't — not really — because you weren't applying the same standard. You were each applying your own reading of a rubric that said something like: "An A paper demonstrates a sophisticated, nuanced argument supported by compelling evidence and marked by clarity and originality of thought."
That sentence sounds rigorous. It is, in practice, a Rorschach test. "Sophisticated" means one thing to the instructor who prizes theoretical framing and another to the one who prizes clean prose. "Compelling evidence" is a judgment call that two experts will make differently in good faith. A holistic rubric — a single paragraph describing the ideal at each grade band — doesn't standardize grading. It hides the disagreement inside adjectives, then lets it surface as inconsistent grades that students, quite reasonably, experience as unfair.
The fix is well known and rarely done: convert the holistic rubric into an analytic one. Break the single judgment into named criteria, define concrete performance levels for each, and write a real descriptor in every cell so graders are pointing at the same evidence when they assign a mark. It's not conceptually hard. It's just tedious enough that it never quite happens before the semester starts. This playbook is about doing it in an afternoon.
Why holistic rubrics drift across sections
A holistic rubric asks each grader to hold the entire quality of a piece of work in their head at once and map it to a band. That's a reasonable thing for a single expert grading their own students to do — it's fast, and their internal standard is at least consistent with itself. It falls apart the moment two people have to agree, for a specific reason: the rubric never externalized the standard. It stayed in each grader's head.
When the standard lives in the grader rather than on the page, three predictable things go wrong across sections:
- Different weightings. One instructor unconsciously weights argument at 60% of the grade; another weights it at 30% and cares more about use of sources. The holistic paragraph never said, so both are "following the rubric" while producing different grades.
- Different thresholds. Where exactly is the line between "good" and "excellent" evidence? The paragraph gestures at it with an adjective. Two graders draw the line in different places, consistently, and neither is wrong by the text.
- Halo effects. A holistic judgment lets one strong dimension — usually fluent writing — inflate the whole score, and it lets one weak dimension tank it. Because the criteria aren't separated, the grader can't see themselves doing it.
None of this is a competence problem or a diligence problem. It's a specification problem. The rubric under-specifies the standard, so each grader fills the gap with their own, and the gaps don't match.
Anatomy of an analytic rubric
An analytic rubric replaces the single paragraph with a grid. Down the side: the distinct criteria you're actually judging. Across the top: performance levels. And — this is the part that does the work — a concrete descriptor in every cell, describing what that criterion looks like at that level.
The transformation of one dimension looks like this:
Holistic (buried in a paragraph): "...supported by compelling evidence..."
Analytic (its own criterion, four levels, every cell filled):
| Criterion | Exemplary | Proficient | Developing | Beginning |
|---|---|---|---|---|
| Use of evidence | Every claim is backed by specific, well-chosen evidence; sources are integrated, not merely cited | Most claims are supported by relevant evidence; integration is uneven in places | Evidence is present but generic, thin, or loosely connected to the claims it supports | Claims are largely unsupported, or evidence is misused or irrelevant |
Now two graders reading the same paper aren't debating what "compelling" means. They're each asking a concrete, answerable question — is every claim backed by specific evidence, or are some claims unsupported? — and landing in the same cell far more often. The descriptor did the standardizing that the adjective couldn't.
The non-negotiable feature is that every cell has a descriptor. A grid with named criteria but empty middle cells — where "Developing" is left blank because it's just "worse than Proficient" — is a holistic rubric wearing a grid's clothing. The whole point is that a grader can read across a row and know exactly where the boundaries fall.
Converting yours in an afternoon
Here's the sequence for turning the paragraph you have into a grid your co-instructors can apply the same way.
- Extract the hidden criteria. Reread your holistic rubric and underline every distinct thing it's actually judging — argument, evidence, organization, mechanics, originality. These buried dimensions become your rows. Most holistic paragraphs contain four to six criteria fused into one sentence.
- Decide the performance levels. Three to five columns. Four is a good default — it avoids a lazy "middle" and forces a real distinction between adequate and strong. Name them plainly (Exemplary / Proficient / Developing / Beginning) so the labels don't smuggle in ambiguity.
- Anchor the top and bottom of each row first. For each criterion, write the "Exemplary" descriptor and the "Beginning" descriptor. These are the easiest — you know what great and poor look like. This defines the range the middle cells have to divide.
- Fill the middle cells — this is the real work. Describe "Proficient" and "Developing" concretely, in terms of observable features of the work, not degree words. "Most claims supported; integration uneven" beats "pretty good evidence." Every middle cell is a place two graders could otherwise disagree, so every middle cell earns its descriptor.
- Norm on real samples. Take two or three papers you've already graded, have each co-instructor score them independently against the new grid, and compare. Where you land in different cells, the descriptor is still ambiguous — fix the wording until you agree. An hour of norming on real work is worth more than any amount of solo polishing.
- Weight the criteria explicitly. Decide what fraction of the grade each row carries, and write it down. This is the "different weightings" failure, solved on paper. Now every grader is weighting argument the same amount, because the rubric says so.
Done honestly, this is an afternoon — an hour to extract and structure, an hour to write descriptors, an hour to norm. And it's an afternoon you spend once, that saves you the reconciliation conversation every grading cycle after.
How the tool fits
The reason this doesn't happen — despite everyone knowing it should — is step 4. Writing a concrete descriptor in every cell of a five-criteria, four-level grid means composing twenty distinct sentences that are consistent in tone, parallel in structure, and genuinely distinct from their neighbors. That's the tedium that kills the project at 4pm.
TeachingsByDesign's rubric builder and rubric warm-up exist to get you past exactly that wall. Paste in your holistic rubric — or start from the outcome the assignment is meant to assess — and the warm-up structures it into criteria by performance levels with a descriptor drafted in every cell. You're no longer facing an empty grid; you're editing a full one. The generated descriptors are a starting point, not gospel: your job becomes reading across each row, sharpening the boundaries to match how your team actually reads work, and cutting anything that doesn't fit your discipline. The tool does the composition; you do the norming and the judgment, which is where a rubric earns its agreement.
Because the rubric lives alongside the assessment and the outcome it measures, the same grid you built for agreement is the one you export clean and hand to every co-instructor, TA, and section leader — so they're all grading against the identical, fully-specified standard, not their own reading of a paragraph.
The bottom line
Multi-section courses grade inconsistently not because the graders are careless but because a holistic rubric hides the standard inside adjectives and lets each grader fill the gap with their own. The fix is an analytic rubric: named criteria, defined performance levels, and a concrete descriptor in every cell, so two graders reading the same work ask the same answerable questions and land in the same place. Extract the buried criteria, anchor the extremes, write the middle cells in terms of observable features, then norm on real samples until you agree. It's an afternoon of work that ends the reconciliation conversation for good — and the only genuinely tedious part, filling every cell, is the part worth handing off.
TeachingsByDesign's rubric warm-up turns a holistic paragraph — or a bare outcome — into a full analytic grid with a descriptor in every cell, ready for you to norm, sharpen, and export clean for the whole teaching team. Grade the same way, every section. See how it works.