Three real datasets. No passive quizzes.
Do the work of an evidence-synthesis team: make a decision before seeing the result, change the study set, challenge a claim, and produce a source-linked decision record.
Source boundaries: Counts and estimates come from publicly documented R teaching datasets. Study subsets and learner-editor scenarios are deliberately constructed for teaching. No invented patient outcomes, religious instruction, causal proof or treatment recommendations. Your score measures participation, not verified research competence.
- 1. Commit
- 2. Predict
- 3. Reanalyse
- 4. Review
- 5. Report
Which inclusion policy can you defend?
Predict what adding the final selected row will do
Reconstruct the evidence yourself
Which scientific claim survives review?
Your evidence verdict, in your own words
Progress is saved locally on this device.
Facilitator use: a reproducible case conference
Ask one learner to defend the precommitted selection rule, another to reconstruct the effect, and a third to challenge the clinical interpretation. Participants submit their own dossier. The source data are real; the peer-review exercise and intervention sequence are instructional, not archived conversations.
For independent R reproductions use magnesium, BCG and lidocaine. Each source's study counts and methods must be checked against the appropriate R package. Peer review, accessibility testing with assistive technology, learning-outcome evaluation and professional translation reviews remain necessary.