Write the rules before seeing the answer.
The real historical case is intravenous magnesium after suspected acute myocardial infarction. Specify what would count as eligible evidence and which endpoint you intend to study. A classroom protocol is not a PROSPERO registration.
Two reviewers. Six real archival trial records.
Run a blinded practice pass using actual `metadat::dat.egger2001` record labels and 2×2 counts. Those counts alone cannot confirm all clinical eligibility criteria; retrieve primary methods for real decisions. Reviewer B will not be shown A's answers until both passes finish. Two different people must participate for genuine human independence; this page cannot authenticate reviewer identities.
Open the complete 16-record R archive ↗ (these six rows are deliberately selected for teaching).
No screening judgement is automatically graded or claimed to be a true systematic-review eligibility verdict.
The in-page A/B practice is not independently authenticated. For two actual humans on separate devices, use the linked Review Exchange: export two independent decision files and adjudicate after both are finished. Human review of the original publications remains necessary.
The year changes. Does the estimate?
The selected magnesium rows can be replayed by publication year: early reports, a subsequent 1989 trial, LIMIT-2 (1992), and the much larger ISIS-4 (1995). Predict before revealing each Peto odds ratio. This is retrospective teaching chronology over six source records—not the systematic reviews actually published in each year.
| Through | Records (k) | Peto OR | 95% CI | Pre-reveal prediction |
|---|
The official full 15/16-trial R example is a different comparison. Use AllMeta for further work, but do not describe these retrospective snapshots as a registered living review.
Your statistical laboratory — real six-study lidocaine data.
The Hine et al. archival data (metadat::dat.hine1989) contain genuine event counts. Select two to six studies and calculate the treatment-minus-control risk difference (null = 0) with a common-effect inverse-variance teaching estimator. This is not the magnesium Peto odds ratio or a clinical recommendation.
Open the same study effects in your AllMeta apps
Input scale: RD on identity scale; choose label “Risk difference” and null 0 in the Workbench or Forest Plot Viewer. Other estimators and variance conventions may yield different pooled results. Compare the methods explicitly.
Privacy/compatibility: direct staging works only when the course and AllMeta share the same GitHub Pages origin. It checks for existing bus data and requires your explicit approval before replacing anything. Otherwise use CSV. Synthesis cannot certify the result displayed by an external app.
The peer-review courtroom.
A demanding reviewer has identified six possible mistakes across screening, statistics, protocol design, evidence interpretation, certainty and reporting. Choose your defence and receive domain-specific feedback. Wrong answers can be retried; errors guide personalised practice but never issue a certificate.
Produce a capstone a methodologist could review.
Explain what the documented archive proves and what it does not. Distinguish selected classroom effects from a full review. Your responses are stored locally and checked for completeness, not scientifically adjudicated. Human review is needed for accreditation.
Progress, accessibility and boundaries
Your educational draft is stored only in this browser. A different person or device cannot automatically see it. Learners may use a keyboard and replay tasks; the course does not enforce timers. Export notes before clearing browser data. No patient-level information is collected.
Source-code and platform test records: Synthesis GitHub, AllMeta GitHub. New English teaching prompts are provisional in all other course-language routes until reviewed by fluent subject experts.