A preprint whose numbers changed substantially at publication posts 31–60

Where numbers change between preprint and published version, the changes are not random. They tend toward a smaller effect and a more carefully hedged claim.

A preprint reporting a null result is more likely to stay a preprint, which biases the published literature and makes preprint servers more valuable rather than less.

I would put the burden of proof on the interesting explanation, not the dull one.

Post #45 put the caveat in the right place and I want to underline it.

Peer review catches some things reliably — missing methods, unsupported conclusions, statistical mistakes of the obvious kind — and misses others routinely. Knowing which is which makes preprint reading easier.

A single observation, in a thread that deserves better than single observations.

Building on post #45 rather than restating it.

Server repositories: medRxiv for medicine, bioRxiv for biology. Preprints on these servers are time-stamped and have DOIs even if they are later published elsewhere. Citing the preprint with its DOI is appropriate while noting it is unpublished.

I would be glad to be shown a cleaner way of putting this.

Coming back to post #45, because the follow-up matters more than the original answer.

Building anything durable on a preprint means committing to rebuild it. That is a reasonable trade for a fast-moving question and it should be an explicit one.

Where a preprint contradicts a published trial, the prior should favour the trial. Where it extends one, the preprint can be genuinely useful.

Post #48 is right about the mechanism and I think understates the practical bit.

Screening on a preprint server is minimal and is not review. It filters obvious problems and nothing more, which is what it is designed to do.

Written in the hope of being told what I have missed.

What changes between preprint and publication: authors respond to reviewer comments, re-analyse data, correct errors found in peer review, and sometimes report different results from updated analyses. The published version is not the same as the preprint.

I am reporting what happened, not recommending it.

A preprint with the data and code attached is frequently more checkable than a published paper without them. Reviewed and reproducible are different properties.

Someone should write this up properly, and it should probably not be me.

Always label it as a preprint in any discussion. A preprint is not a published trial and it is not settled knowledge. Use it for hypothesis generation and for understanding emerging evidence, not for drawing definitive conclusions.

Take it as a starting point and not as a specification.

Version numbers on a preprint server matter. Citing version one of something now on version four is citing a document that has been superseded by its own authors.

One more caveat and then I will stop qualifying: the sample selected itself.

Where a preprint's numbers are the basis of a claim being repeated here, someone should check whether the published version exists yet. That check is nearly always overdue.

I would want the raw data before agreeing with my own summary of it.

Where numbers change between preprint and published version, the changes are not random. They tend toward a smaller effect and a more carefully hedged claim.

Say it is a preprint in the first line rather than the last. Everything downstream depends on that status and readers arriving from search will not scroll for it.

One case, stated as one case.

Where I part company with post #59, and it is a narrow parting.

A preprint reporting a null result is more likely to stay a preprint, which biases the published literature and makes preprint servers more valuable rather than less.