What Would Contradict Your Claim? Decide Before You Consult the Data.
Learning Analytics Interpreter reads your pilot data and the claim you're tempted to make from it — but only after you write down what result would have told you the claim was wrong. Then it hands back a six-field Interpretation & Revision Brief, not a verdict.
Bring your data
What did the pilot produce, and what are you tempted to say about it?
Paste completion stats, an assessment-score summary, a short survey excerpt, or an LMS export — whatever you have. This also handles multimodal evidence: a video-coded engagement rating, a screen-capture debrief note, eye-tracking or physiological summary data, or a description of what an AI-assisted process trace showed. No file upload; a plain-text paste or description is enough either way.
Before pasting: remove names, IDs, and any single-learner-identifiable rows. Paste summary numbers, not a roster — this tool does not scan for or scrub personal information automatically.
No pilot yet? You can still rehearse this tool with numbers you anticipate getting, to pressure-test your evaluation plan ahead of time. It stays clearly labeled as hypothetical throughout, never treated as a real result, and you can run this again with real data once you have it.
Example of a clean paste: “Completion: 42/50 (84%). Post-test mean: 78% (n=42), pre-test mean: 61%. Survey (n=38): 4.2/5 satisfaction.” A short survey excerpt or LMS export snippet works too. Multimodal example: “Screen-capture debrief coding (n=20 sessions): 12 learners self-corrected without prompting, 8 needed the AI copilot to point out the error. Average time-to-first-correct-attempt: 4.2 min.”
0/6000
What the plan intended to measure, and any pre-committed revision criteria — e.g. “if completion drops below 70%, redesign the pacing.” If you don't have one, the Brief will say so and give generic revision guidance instead of criteria-matched guidance.
0/2000
One per line — e.g. “Completion went up in the redesigned version” or “Learners in Group B struggled more.”
0/1000
First, say whether you have real data yet or want to rehearse with anticipated data.