This is a working draft for Leslie's review. All three demonstrations are scripted; edit the question ladders and example exchanges freely. Each demonstration is scored against its published rubric — this page is how to run it, not the rubric itself.
The demonstration is the part of the course that cannot be faked, outsourced, or generated. A student stands in front of you and shows understanding in real time, against a rubric, out loud. Your job is to run it the same way every time: a setup that makes faking impossible, a question ladder that probes past rehearsal, and a clean binary verdict recorded against the rubric.
The three rules that govern every demonstration
- The student does the work; you only ask. No coaching, no leading questions, no "are you sure?" that signals the answer. You probe understanding — you never supply it.
- The verdict is binary. Mastered or not yet. You are not assigning a score in the room; you are deciding whether the published rubric's bar was cleared.
- A "not yet" is a checkpoint, not a failure. How you deliver it determines whether the student comes back. Name the specific gap, give the re-attempt path, and keep the door open.
Every demonstration below climbs the same shape: from recall at the bottom to a rehearsal-proof rung at the top. Keep climbing until you reach the student’s real ceiling — a memorized answer falls off near the top.
Demonstration 1 — Experiment-design defense
Scored against the experiment-design defense rubric · after Unit 04.
Setup. The student designs, runs, and defends a controlled experiment of their own — say, does raising a ramp make a toy car roll farther? They state the question, name the one thing they'll change and the things they'll hold constant, run at least two conditions, and record the distances in their notebook while you watch. Because it's their design and their data, there is nothing to outsource.
The question ladder:
- Design: "Tell me your question, then show me your fair test — what changes, what stays the same?" (Can they isolate a single variable?)
- Procedure: "Run it and record your results." (Can they take clean, repeatable measurements — and repeat a condition?)
- Reasoning: "What does your data let you conclude — and what does it not?" (Do they reason from the evidence, not from what they expected?)
- Counterfactual: "If the car had rolled farther from the lower ramp, what would you check first?" (This is the rehearsal-proof rung — a memorized answer dies here.)
Demonstration 2 — Timed data interpretation
Scored against the timed data interpretation rubric · after Unit 05.
Setup. Under a time limit, the student is handed an unfamiliar dataset — say, how long ice cubes take to melt at several room temperatures — and must table it, graph it, and draw a justified conclusion. The clock is the integrity mechanism: there's no time to look anything up, and reading a real dataset honestly can't be generated.
The question ladder:
- Table & plot: "Set this up as a table, then graph it — tell me what goes on each axis." (The core skill: turning raw numbers into a readable graph.)
- Interpret: "What's the pattern, and which points tell you?" (Can they justify the trend from the data in front of them, not from a memorized example?)
- Discriminate: "Is that a real trend, or could it be noise? How would you tell?" (This is where a shaky student reveals themselves.)
Demonstration 3 — Oral lab-notebook defense
Scored against the lab-notebook-defense rubric · end of each unit.
Setup. The student walks you through their own notebook: the question, the method, the data, the anomalies, the interpretation. Because it is their recorded work, the defense is unfakeable — a student who didn't do the thinking can't narrate the decisions behind it.
The question ladder:
- Narrate: "Walk me through what you did and why." (Does the story hold together?)
- Anomaly: "This data point doesn't fit — what happened?" (Did they notice, and can they reason about it honestly rather than hide it?)
- Method critique: "What would you change if you ran it again?" (Do they understand the method's limits and sources of error?)
- Transfer: "How does this connect to the concept from this unit?" (Integration — scored on its own line.)
Reading the room: mastered vs. approaching
| Mastered | Approaching (“not yet”) |
|---|---|
| Answers the counterfactual rung without hesitation — reasons forward from understanding. | Strong on recall rungs, collapses at the counterfactual. The understanding is memorized, not built. |
| Owns anomalies and sources of error, and reasons about them. | Hides or hand-waves the data that doesn't fit. |
| Connects the observation to the mechanism and the lab to the concept unprompted. | Knows the test in isolation; can't trace the reasoning behind it. |
Recording the verdict
Mark the rubric in the room while it's fresh — pass or not-yet on each rubric line, with a one-line note on the deciding moment. If the verdict is "not yet," that note becomes the re-attempt instruction. Do not soften the rubric to avoid an awkward conversation; soften the delivery instead.
Delivering a "not yet" so the re-attempt stays likely
- Lead with what was solid before naming the gap — the student needs to know the work wasn't wasted.
- Name one specific thing to close, not a list. ("The counterfactual on the ramp experiment — come back when you can reason through how a lower ramp could still give the farther roll.")
- Give the date and the path. A "not yet" with a re-attempt slot on the calendar is a checkpoint; a "not yet" with no path is a wall.
- Coach or hint during the demonstration — including tone, eyebrows, and "hmm."
- Ask leading questions that contain the answer.
- Let partial credit creep in. There is no 7-out-of-10 here — the bar is cleared or it isn't.
- Move the bar for a likeable student, a tired afternoon, or a parent in the room.