Timed prediction-and-test
The student is handed an apparatus and a clock. Working against time, they measure what they need — a launch speed, a ramp height, a cart's mass — and predict what will happen: exactly where the projectile lands, or how fast the carts move after they collide. Then they run it live, once, and compare the prediction to the result. There is nothing to copy and no key to consult: the launch is real, the time is real, and the number they wrote down has to be defended against the mark the ball actually left.
| Criterion | Developing | Proficient | Mastery |
|---|---|---|---|
| Prediction method | Reaches for the wrong physics model, or skips the measurements the prediction needs. | Uses the right model but takes a sloppy measurement, or sets the calculation up with a gap. | Chooses the correct physics model, takes the measurements it needs, and sets up the calculation cleanly to a definite predicted value. |
| Execution under time | Runs out of time, or fumbles the test so badly the result is unusable. | Completes the test within the limit but rushes a step that adds avoidable error. | Runs the test cleanly and in control, finishing well within the time limit. |
| Accounting for the discrepancy | Ignores the gap between prediction and result, or blames it on “human error” with nothing behind the phrase. | Notes the discrepancy and names a source or two but cannot say which way each one pushed the result. | Compares prediction to result honestly and explains the real error sources — air resistance, a mismeasured height — and their direction. |
| Communication of reasoning | Cannot connect the claim to the model or recorded evidence. | Explains some choices but leaves an important link unsupported. | Explains the model, evidence, and a limitation using the agreed communication method and accommodations; composure is not a science criterion. |
Integration is reported separately and cannot lower the science grade or block a science demonstration pass. Science and practical criteria determine that pass. Before assessment, publish the target list, task, permitted references, timing, and accommodations. Assess scientific reasoning and safe technique, not confidence or speaking speed.
“I measured the launch speed off the ramp and the table height, so this is a projectile problem — horizontal and vertical run independently. I predicted it lands 1.42 meters out. It hit at 1.37, a little short, which is what air resistance would do to a light ball, so I’m confident the model is right and the gap is real.”
“I thought it would land about there, but it landed somewhere else. I ran out of time to check my numbers, so I’m not sure why it missed.”
This assessment is AI-proof by design: it happens at the bench, with a real launch and a real clock. No chatbot can measure the ramp, commit to a landing point, and then stand behind the number when the ball hits the floor. The setup differs from student to student, so there is no answer to look up — mastery is shown by predicting, testing, and accounting for the gap in person, not by submitting.