Nutrition-analysis defense
This is a live exam at the bench. The student pulls real numbers — a day of their own meals, or a single food’s nutrition label — and works out what the data actually says: where the calories and macronutrients come from, and which nutrients run high or low against the guidelines. Then the guide starts asking: why read the serving size that way, what does the evidence actually support, what one change would you reason toward and why. There is no worksheet to copy and no answer to look up: the student stands over their own data and defends the reasoning out loud.
| Criterion | Not yet | Approaching | Mastered |
|---|---|---|---|
| Gathering & reading the data | Misreads the label or the record — wrong serving size, skips added sugar, cannot pull the numbers cleanly. | Reads most values but confuses one serving with the whole package, or misses a key nutrient. | Reads the nutrition data accurately — serving size, calories, macronutrients, added sugar — and pulls a complete, honest set of numbers to work from. |
| Analysis & reasoning from evidence | Reports numbers without interpreting them, or draws a conclusion the data does not support. | Spots one pattern in the data but cannot connect it to what the body needs. | Interprets the numbers against evidence-based guidelines — where energy and nutrients come from, what runs high or low — and reasons from the data rather than from assumptions. |
| Evidence-based recommendation | Offers a fad rule or a “good food / bad food” verdict the data does not support. | Suggests a change but frames it as an order rather than as reasoning, or cannot tie it to the evidence. | Frames a recommendation as reasoning — “the evidence suggests…” — grounded in the data and the guidelines, without prescribing calorie targets or judging the food as good or bad. |
| Defending under questioning | Folds at the first follow-up or recites a memorized line that does not fit the data. | Answers some follow-ups, falters when asked to justify a number or a choice. | Handles unrehearsed follow-ups about this analysis with sound, on-the-spot reasoning from the evidence. |
| Integration (cross-domain) | Treats the analysis as isolated numbers; makes no cross-domain connection. | Names a link to history, reading, or statistics but cannot defend why it matters. | Connects the work to evidence-based reasoning — including James Lind’s scurvy trial as the root of testing a claim against data — and defends why the connection matters. |
“Most of the added sugar in this day comes from two drinks, not the meals — the labels show forty grams there, well above the guideline. The evidence suggests swapping one drink would bring the added sugar closer to that guideline, and I can point to the exact numbers I’m reasoning from.”
“This food is bad because it has sugar in it, so I’d just cut it out. I’m not really sure what the numbers say or how much is in a serving.”
This assessment is AI-proof by design: it happens at the bench, with the student’s own real data and a live conversation. No chatbot can gather a day of real meals, reason from numbers it did not pull, or hold up under a follow-up question about a recommendation it cannot defend. Mastery is shown by analyzing and defending — not by submitting.