How Many Performance Levels Should a Rubric Have?

The number of columns on a rubric looks like a minor formatting choice, but it changes how reliably the rubric scores work and how much a student learns from the result. Too few levels and you can’t tell a strong paper from an exceptional one; too many and two graders start disagreeing about which cell a paper belongs in, because the wording separating adjacent levels gets thinner the more levels you add. There’s no single right number, but there is a reliable way to pick one: start from what the score is for, then choose the smallest scale that lets you say what actually needs saying.

What performance levels are

Performance levels are the columns of an analytic rubric (or the discrete steps of a holistic one) — the set of quality bands a piece of work gets sorted into, each with its own descriptor. Four levels running from, say, Beginning to Exemplary is a common default, but the levels themselves aren’t the point; the descriptors attached to them are. A scale is only as useful as the wording that separates one level from its neighbor, which is why adding levels without adding real distinctions between them makes a rubric worse, not more precise.

The case for four levels

Four levels is the most common choice in K–12 rubrics, and for a specific reason: it removes a safe middle option. With an odd number of levels, a grader unsure how to score a borderline paper can always retreat to the exact center, and over a whole stack that middle column quietly absorbs papers that actually deserved a real judgment call either way. Four levels forces every paper to land on the “meets or exceeds” side or the “not yet” side, which produces more decisive, more useful feedback and — because there’s nowhere to hide indecision — often a more consistent scale across a whole class set.

Odd vs. even numbers (the fence-sitting problem)

The four-level case above is really an argument about odd versus even, not about the number four specifically. Any even scale (two, four, six levels) shares the same property: no exact middle, so every score is a real lean in one direction. Any odd scale (three, five levels) has a true center value, which is easier to default to under time pressure and harder to defend when a grade is questioned — “I wasn’t sure, so I put it in the middle” is not a sentence you want to have to say to a parent. That doesn’t make odd scales wrong; a genuine middle category is sometimes exactly what you want to express (see the three-level case below) — but choose it because the middle category means something specific, not because it’s the safe default when you can’t decide.

Three levels for quick, formative checks

A three-level scale — something like Not Yet, Approaching, Meets — is often the right size for a quick formative check: an exit ticket, a rough draft, a homework spot-check where the goal is a fast signal of who needs more support, not a finely graded score that will end up in the gradebook. Three levels are fast to score and fast for a student to read at a glance, and the coarser resolution is a feature here, not a limitation — you’re optimizing for speed and clarity over precision because the stakes of the assessment don’t call for finer distinctions yet.

Five to six levels for high-stakes work

On the other end, a major summative assessment — a final project, an end-of-unit essay, anything used for a report-card grade or a high-stakes decision — can justify five or six levels, because the additional resolution is worth the extra time it takes to write and apply. The catch is that each additional level needs its own genuinely distinct descriptor; simply inserting a level between “proficient” and “exemplary” without writing new, specific language for it just gives graders more cells to argue about, not more information. If you can’t write a sentence that says what separates level 4 from level 5 that a colleague would agree with, you don’t have five real levels yet — you have four levels and one duplicate.

Labeling levels

How you name the levels matters almost as much as how many there are. A common mistake is labeling only the bottom of the scale in specific, often deficit-focused language (“Missing,” “Incomplete,” “Poor”) while the top levels get vaguer, more generic praise words (“Good,” “Great”) that don’t actually describe anything. Susan Brookhart’s guidance on rubric design pushes the opposite approach: label every level, including the top ones, in terms of what’s actually present in the work — “sustains a clear argument throughout” communicates a target students can aim for, where “excellent” only tells them they succeeded, not at what. Whatever words you put in the header row, make sure a student reading only the label (not yet the full descriptor) still gets a real signal of what to work toward.

Aligning levels to points and grades

Once the scale is set, decide how it turns into a number before you need the conversion mid-grading. A four-level scale often maps directly to four point values (4, 3, 2, 1) that sum into a total, or converts to a percentage band per level if your gradebook expects one. Whichever mapping you use, keep the spacing between levels meaningful — if a “3” and a “4” are one point apart but represent a much bigger quality gap than a “2” and a “3,” the numbers are quietly lying about how different those two papers actually are. If you’re reporting against fixed proficiency levels rather than a single assignment’s point total, our guide to standards-based grading with rubrics covers how that reporting layer is usually built.

Keeping scoring consistent

Whatever scale you land on, consistency across a stack of papers depends more on the quality of the descriptors than on the count of levels — a well-written three-level scale will out-perform a sloppy six-level one every time. Pilot a new scale on two or three real or representative samples before using it on a whole class, and if a sample makes you hesitate between two adjacent levels, that’s the signal the wording separating them needs to be sharper, not a sign you need to add a level in between. Our step-by-step guide to writing a rubric covers that piloting process, and how to write descriptors precise enough to hold up under it, in more depth. If you want a working scale to start from rather than building one from scratch, our free blank rubric template ships with a four-level scale already in place, ready to relabel or resize to fit your assignment.