Review: an evidence digest and the citation check that delayed it

On the subject in the title: Review: an evidence digest and the citation check that delayed it Working notes rather than a conclusion.

I was wrong about evidence digest in a thread last spring and I would like to correct it publicly rather than quietly.

The error was in the units, which changed the conclusion by an order of magnitude. Setting out the corrected version, and the way I now check for that class of mistake.

The limitations section is the part most often thin at review and most often load-bearing afterwards.

Building on post #2 rather than restating it.

Review of promoted topics: when a topic is nominated for promotion to a maintained documentation page, the draft is posted here. Community review happens before promotion. Objections are welcome.

For what it is worth, the same held on the two occasions I checked.

Nothing in a maintained document is medical advice and the review should confirm the page says so where it matters rather than only in a footer.

Adding it in case it saves somebody the afternoon it cost me.

A review date is a commitment rather than a formality, and a shorter cycle for a fast-moving subject is the whole point of setting it deliberately.

Answering the question post #4 raises rather than the one it answers.

Where a review raises something the author disagrees with, recording the disagreement in the review thread is better than resolving it privately.

Where content is lifted from a topic, attribution should be in the document rather than only in the review thread.

Happy to be corrected if someone holds better data than mine.

Check the links. Promotion frequently breaks a cross-reference because the source topic now points at the document rather than the reverse.

Adding the measurement that post #6 says would settle it.

Check that each claim is sourced or explicitly labelled as community practice. Unmarked practice claims read as sourced ones.

I would be interested in a counterexample if anyone has one.

Post #8 describes the usual case. This is about the unusual one.

Check the arithmetic in every worked example by hand. A wrong worked example is worse than no example and survives longer.

That holds for the case as described. Change the assumptions and it may not.

Ask what the page would look like if it were wrong. A document with no stated failure conditions is hard to review and harder to maintain.

It is one reading of the data and not the only reasonable one.

Building on post #11 rather than restating it.

Review the scope as well as the content. A page attempting too much is the commonest structural problem at this stage.

The most useful review comment names a sentence and says what is wrong with it. General approval and general concern are both hard to act on.