Is the proficiency decision consistent?
Two questions every school using proficiency-based or mastery-based grading needs to answer: whether the decision is consistent across teachers and across the year, and whether support before a resubmission is being counted the same way. These two tools answer both questions on paper.
Both are free, licensed CC BY 4.0, and work without an account, without QLM software, and without student data leaving the school. Adapt them to your scales and your grading policy. The attribution line goes wherever your district places third-party content credits: a footnote, an appendix, or a references section.
Norming Calculator
Type the grid instead of writing it. The calculator computes exact and adjacent agreement, keeps a running log in your browser, and exports to CSV. Nothing is sent anywhere. It works offline after the first load.
Norming Protocol with a Scoring Sheet
Teacher teams that grade on proficiency scales already norm on student work. This protocol keeps that practice and adds a record. A session takes about 40 minutes and ends with two agreement percentages written down with the date, so a team can see whether its consistency held from October to May. The numbers describe the scale wording, and the log never records which scorer gave which score.
Reading your number
The number is only meaningful against your own team's later number on the same target and scale. A rise in exact agreement over several sessions is consistent with clearer scale wording, but check whether the sample tasks were comparable in difficulty before drawing that conclusion. If exact agreement stays flat while adjacent agreement is high, the team agrees within one level but cannot distinguish two adjacent levels. The descriptor between them is the one to rewrite. If both are flat, look at the one-decision column. The same descriptor rewritten three months running usually needs a structural change, not another tweak.
What you get: A two-page PDF with the eight-step protocol, a worked example, a blank scoring grid, and a running log. The DOCX is the same file for editing.
Time: About 40 minutes per session. Once a month per learning target works well.
Cite as
Quantum Learning Machines. (2026). Norming Protocol with a Scoring Sheet (v1.0). Zenodo. https://doi.org/10.5281/zenodo.22827274
File checksums
Revision Credit Rule Template
Retakes and revisions are how students reach proficiency. What varies from classroom to classroom is how the support a student received before resubmitting counts toward the decision. This template puts that rule on one page, in a form students can read and colleagues can share, with a twice-yearly check on whether students marked proficient held the level on their next unsupported demonstration.
Research backing
This template is the school-facing version of the SDA-001 finding: in a preregistered analysis of 12,716 students and 985,813 scored events, crediting assisted work inflated measured mastery by 21 percentage points. The rule that reads only first attempts predicted later unaided performance best (AUC 0.658, against a per-skill constant at 0.595). Read the preprint (EdArXiv, September 2026)
What you get: A one-page PDF template with eight sections, including a grid of eight support types and three ways each can count. The DOCX is the same file for editing.
Time: One team meeting to write the rule for a target. Ten minutes twice a year to check it.
Cite as
Quantum Learning Machines. (2026). Revision Credit Rule Template (v1.0). Zenodo. https://doi.org/10.5281/zenodo.22827286
File checksums
Short answers
Does this require QLM software?
No. Both tools work on paper.
Does any data leave the school?
No. The scoring sheet records levels, and the log records team numbers. Neither records names.
Can we change the wording?
Yes. CC BY 4.0 lets you adapt and redistribute. The attribution line goes wherever your district places third-party content credits: a footnote, an appendix, or a references section. The line is: "Adapted from [Tool Name] v1.0, Quantum Learning Machines, 2026. Licensed CC BY 4.0."
Where do these come from?
They grew out of conversations in 2026 with state and district networks implementing proficiency-based and mastery-based learning. The measurement questions they answer, whether a proficiency decision is consistent and how support before a demonstration should count, are the same ones YardStick is built around. The revision credit rule template is grounded in a preregistered analysis of 12,716 students showing that crediting assisted work inflates measured mastery. Preprint (EdArXiv, September 2026).
Do these work for mastery-based learning?
Yes. The protocol works with any ordered proficiency or mastery scale of 2 to 6 levels. The credit rule template uses the same support types whether the school calls it proficiency-based, mastery-based, or competency-based grading.
Used these with a team?
Send your two percentages, the date, and the target name to hello@quantumlearningmachines.com. No scorer names, no grid, no student data. Just the two numbers.
We use them in aggregate only, never school-identified, to learn whether the protocol works in practice. If we ever report aggregate statistics publicly, we will say so and link them here. A team can ask us to delete their numbers at any time. This is not a sales conversation.