A ninth-grade writing rubric separates a “4” from a “3” with language like “demonstrates sophisticated control of voice” versus “demonstrates control of voice.” Ask the teacher who wrote it what “sophisticated” means that “control” alone does not, and most give the same answer: I know it when I see it. Rubrics are supposed to remove that kind of judgment call from grading. Instead, too often, the teacher moved the judgment call into a single adjective and called the problem solved.

The fix is not a more detailed rubric. Adding criteria multiplies vague terms instead of replacing them. A five-column rubric with the same adjectives repeated at different intensities is still a subjective adjective problem, just a longer one. The actual fix is rewriting performance levels around actions a student can take, and a reader can point to in the text, not qualities a teacher recognizes only after the paper is finished. This is the hardest part of writing a usable rubric—teachers must understand exactly what they are looking for in a grade ahead of time and be willing to live up to it. Students will push back when a teacher fails to follow through on a promised grade.

Bob Broad documented this gap directly. In What We Really Value: Beyond Rubrics in Teaching and Assessing Writing (Utah State University Press, 2003), Broad asked experienced writing faculty to explain, in their own words, why a paper earned the score it did, then compared those explanations to the rubric language meant to standardize the same scores. Unfortunately, the two rarely lined up. Faculty pointed to specific moves in the text: a claim that got narrower instead of broader, a transition that did real work instead of signaling sequence. The rubrics offered adjectives like “sophisticated” and “effective” that no two readers applied the same way. This is where specific wins over vague generalities.

Susan Brookhart reached a related conclusion from the classroom-assessment side rather than the research side. In How to Create and Use Rubrics for Formative Assessment and Grading (ASCD, 2013), she warns specifically against performance levels that differ only by frequency words, such as “always,” “usually,” and “sometimes,” attached to an otherwise identical trait description. A scale built that way asks a student to guess at a frequency rather than identify a missing or present move in their own draft. It also makes the rubric nearly useless as a revision tool, since “sometimes uses varied sentence structure” gives a student nothing to act on between drafts.

When done correctly, the difference shows up immediately once a descriptor gets rewritten. “Uses sophisticated transitions between paragraphs” becomes “each paragraph’s opening sentence names a specific claim or detail from the paragraph before it.” “Demonstrates strong command of conventions” becomes “fewer than one error per page in subject-verb agreement, comma splices, and apostrophe use.” The first version of each pair asks a teacher to recognize a quality. The second asks a student to check one concrete thing against their own page. Only the second version is checkable before the paper is turned in, which is the only point in the process where checking still changes the outcome.

George Hillocks Jr.’s research synthesis on written composition instruction reached a related conclusion decades earlier: students revise more effectively when feedback names a specific, repeatable craft move instead of a general quality judgment (Hillocks, 1986). A rubric is feedback delivered in advance of the draft instead of after it. The same standard applies: name the move, not the quality.

None of this is free. Rewriting a rubric around observable, checkable actions takes longer than copying last year’s adjectives into a new template, and it requires someone on staff to decide what “sophisticated” is actually supposed to mean in practice. That cost is paid once per rubric. It is paid back every time a student uses the rewritten version to revise a draft instead of waiting for a grade to find out what “good” meant this time.

Further reading