Am I allowed to duplicate results from my own arXiv paper in a journal publication?

Question

I had put up a paper on arxiv about 4 months back, which we had planned on submitting to a journal. However, before I submitted, we got some good reviews and ideas from a colleague at my university. All of the original results were correct and accurate, but now I would like to add some additional simulation results, a more detailed analysis, and also give an authorship to the colleague. However, in this time span, my arxiv paper has been cited a couple of times.

My questions are:

1. I would like to change the title of the paper and do it from fresh, and submit the newer version to the journal. In this case, should I just cite my arxiv article for all the background work, and just present the new contributions (which would be only 1-2 pages); or can I copy over all the content from my original arxiv article?

2. If the latter is possible, how do I indicate to others that they should cite the newer article and not the original one?

Field of research: At the interface of applied mathematics, engineering and computer sciences. Paper likely to be submitted to a SIAM journal.

Answer 1

Manuscripts in the arXiv are permanent and citable, but do not count as peer-reviewed publication. Therefore, there is no inherent problem either with using your arXiv work as part (or all) of a manuscript submitted to peer review or with citing your arXiv work as background on a new work that builds upon it.

If you re-use your arXiv material in a manuscript, however, there are two additional steps that you must take:

1. You must make the source of the material clear, in both the cover letter and the document itself. I personally like verbiage such as:

   "Note: portions of this manuscript have been adapted from [cite]"

2. You need to check the policies of the venue where you submit it. In some fields, such as computer science, most journals do not consider non-peer-reviewed venues such as arXiv prior publication, and so including such material is fine. In other fields, such as biology, most journals do consider them pre-publication, and will consider it self-plagiarism. These policies vary journal-by-journal, however, so you do need to check the individual policy.

Answer 2

In this answer, I'm assuming that the journals you are considering are not prejudiced against publishing preprints. (If this is not the case, I think you may have hard time publishing the content of the already-uploaded preprint, but this is likely largely dependent on the policy of individual editors and journals.)

If the journal is okay with regards to publishing preprints on arXiv, I don't see why they would mind having a "partial" preprint over there (as long as the "preprint" authors form a subset of the "postprint" authors, of course). But I want to say a bit more about your options with regard to arXiv.

I think one think you can do is to simply update the paper you have uploaded to arXiv, along with the added author and possibly other acknowledgements, and then send it to your journal of choice, and (simultaneously) simply replace the arXiv version with this new, updated one. I think this could be a good idea, particularly if the structure of the "old" part is left mostly untouched and you just add more content at the end (and maybe change the abstract and introductory parts a little bit). By your comments about the changes being only one or two pages, I imagine that this is the case here.

If the upgrade will inevitably disrupt the structure of the already-written part, then I think you should follow the advice on this thread (actually, you should probably see that thread either way).

I'm reading between the lines here, but I think you're thinking that what you don't consider replacing the arXiv article is because you want to change the title and add an author or because the preprint has already been cited. But you should know that:

1. You can add an author to an existing paper as well as change the title when replacing a paper.
2. arXiv keeps all versions of uploaded preprints with timestamps, so when someone follows a reference to (an old version of) your paper, they should be able to tell by the timestamp that the reference was to a previous version, even more so if the title has been changed, and if the structure (especially numbering) of the old part is left virtually untouched, it might not matter anyway.