How the rubrics work.

Every rubric in this pack is built the same way, on the same few rules. Read this once and the eight unit rubrics and three demonstration rubrics all make sense — and grading a student stops being a guess.

Three operating principles

These three rules hold across every rubric. They are what make the bar identical for every student and legible to every guide.

Principle 01

We grade decisions, not points.

Each row is a single judgment: Not yet, Approaching, or Mastered. There is no partial credit to haggle over and no points to total. A student has either shown the skill or has not yet — and “not yet” is an invitation to come back, not a verdict.

Principle 02

The vocabulary is published in advance.

Nothing on a rubric is a surprise. The skills, the terms, and the standard for Mastered are all printed before the unit begins. A student can read exactly what they will be asked to do and defend — so the work is to learn it, not to guess what the grader wants.

Principle 03

Three tokens absorb a bad day.

Everyone has an off day — a cold, a missed night’s sleep, a circuit that won’t close on the day. Each student carries three tokens across the course: a token lets them re-attempt one demonstration, no questions asked. Mastery is the goal; one bad morning should never decide it.

From mastery to a grade

Most families still need a letter grade for a transcript. We get there with specifications grading: instead of averaging points, each letter is a bundle — a contract that names exactly what a student must have mastered to earn it. The bundles are cumulative, so the next grade up is always a short, concrete list away.

Mastery

All B-level criteria
All 8 units reached Mastered
All three demonstrations passed in person
Lab notebook: passed every week

Proficient

All C-level criteria
At least 6 of 8 units Mastered (rest Approaching)
2 of 3 demonstrations passed

Foundation

All 8 units at least Approaching
At least 1 demonstration passed
Lab notebook: passed every week (one make-up allowed)

Engagement

Took part and kept a lab notebook on at least half the weeks
Attempted all three demonstrations
Most units still Not yet

What this structure does. A student with a B has not skipped notebook weeks or failed every demonstration — each letter is a promise about what the student can actually do. The grading conversation becomes “here are the two items that would move you to a B” instead of “your average is 79.4%.” The D bundle is the honest “engaged, not yet at standard” contract: it credits the work the student put in while being clear the bar is not yet met. Families who keep a mastery transcript instead of letters can read the same four columns as tiers — Mastery, Proficient, Foundation, Engaged.

What these rubrics are not

Naming what we deliberately left out is the fastest way to understand the design.

Not a points average. A grade is a bundle of mastered skills, never a weighted sum that lets strong essays paper over a skill never learned.
Not curved. The bar does not move with the cohort. Every student is measured against the same printed standard, not against each other.
Not a timed multiple-choice test. The decisive assessments are demonstrated and defended out loud at the bench — the part no chatbot can do for a student.
Not a one-shot. “Not yet” is a checkpoint. Re-attempts and the three tokens exist precisely so a single bad run never freezes a grade.

Why this works

Specifications grading, mastery bundles, and published criteria are not house inventions — they sit on a well-documented body of assessment research. Linda Nilson’s Specifications Grading (2015) makes the case for grades as contracts; Royce Sadler’s work on criteria and standards shows why a published standard outperforms a point total; and the wider rubric literature (Jonsson & Svingby, 2007) finds that explicit, shared criteria make grading both more reliable and more useful to the learner. We simply tuned that scholarship for a hands-on, defend-it-in-person physical science bench.