Analyzing article dates

[OmnesRes]

I think it would be really interesting to look at license preferences over time, but dates with preprints are tricky.

The date in PrePubMed is not necessarily the date of the first version of the preprint. PrePubMed began indexing articles in April 2016. For any article added to bioRxiv after that date the date in PrePubMed should be the date of the first version of the preprint. For articles before this date PrePubMed indexed the version that was newest, not the earliest possible version. As a result, someone theoretically could have submitted a preprint in 2014 and chosen their license in 2014, but then updated their article in February 2016, and PrePubMed will have the date of February 2016. It is possible to reindex the dates to get the date of the first version, but this is assuming that users do not change the license when they upload a new version.

[dhimmel]

It is possible to reindex the dates to get the date of the first version, but this is assuming that users do not change the license when they upload a new version.

Just played around with bioRxiv and confirming that it's possible to change the license when doing a revision.

The date in PrePubMed is not necessarily the date of the first version of the preprint.

This is a caveat that I'm not too worried about. Do you have a sense of how common revisions are? And whether revisions tend to be close in time to each other or years apart?

[OmnesRes]

I haven't been recording revisions, but when I do my periodic eye checks of the preprints it does seem around half of the articles on the front page of bioRxiv will be revisions and not new preprints. I think revisions only occur up to time of acceptance of the preprint to the journal, which could be a span of a few months.
When you visit the most recent version of a preprint with the doi the full history of revisions can be clearly observed in the .article-info page. So if you wanted to do an analysis on how often revisions occur that would also be possible.