# Bag-to-Number Probability Lab — analytics map

| Evidence | What it can indicate | Teacher use |
|---|---|---|
| Favourable outcomes selected for A, B and C | Whether the learner can construct and classify the 12-outcome sample space | Compare probability answers with the underlying event set |
| First and final submitted probability | Initial reasoning and any conceptual change | Distinguish a fixed error from a productive revision |
| Checked attempts and correctness | Outcome and persistence | Separate unfinished work from incorrect work |
| Confidence (1–3) | Certainty attached to the answer | Notice confident misconceptions or low confidence despite success |
| Hint use | Scaffolding consulted | Plan follow-up support |
| Misconception code, explanation and teaching move | A response-supported reasoning gap | Group and address errors using the relevant representation |
| Bounded semantic action sequence | The order of selections, checks, hints, revisions and reset actions | Reconstruct the evidence behind the final result |

Scoring comes only from the three submitted probability answers: A = 1/3, B = 1/2 and C = 0. The selected outcomes are unmarked diagnostic evidence. The final xAPI state contains score/max, quiz.items, hiddenMarks.items, details.timeline, summary, a bounded top-level history, and a visual HTML teacher report in feedback.

No completed 0/3 state is sent at launch. Meaningful evidence is kept locally during work, an explicitly labelled in-progress state is sent only on pause/exit, and the complete report is sent on the learner’s Finish action.

Privacy: no raw keystrokes, authentication values, learner identity copies or unrelated browser data are stored. Submitted answer values are recorded only when committed or checked.

Interpretation limit: the analytics describe behaviour in this interactive and support teacher noticing; they are not a definitive judgment of ability or intent.