I want to cite a paper which was published online (Open Access) in 2012 and later published in a printed journal issue in 2013. Which publication year should be used in the citation?
As StrongBad said, the general rule is to follow the guidelines outlines by the publication venue.
If there is none, then as a rule of thumb, cite the journal version. The reason depends on the type of online publication:
- When the paper was an early access version of the later journal paper and thus identical, the journal paper in some sense replace the online access version.
- When the 2012 paper was posted at arXiv.org or the like, then the later journal version went through peer reviewing, whereas the earlier version did not. So the journal version can easily be more mature.
Note that one role of citations is to assign academic credit. Credit towards a paper that appeared in a reputable journal counts a bit more, so citing the journal version is also a courtesy to the author(s).
If for some reason, for your paper, it is of importance who had an idea first (e.g., when you give a chronological overview of related work), you can just add a note like "An earlier version of the paper appeared in XYZ in 2012" to the bibliography entry of the journal version to make that clear.
Each article, to be properly indexed, is assigned some minimum information:
Journal name given in a correct way. However, this can be 3 of more different things for one journal, e.g. "JAir" = "J Aircraft" = "Journal Aircraft" (randomly chosen journal that has 3 proper names); you can choose any, but usually the middle form is prefered (i.e., basic abbreviations, but not the shortest ones, those are known as "astro. abbrv.")
Journal volume. Starts at one when the journal is introduced, and counts one by one in a way that is basically decided by the publisher, with some rules.
(a) Article's first page, or (b) in cases of per-article page numbering, article code, or (c) in cases of per-issue/number page numbering, number of the issue in the volume (usually called just "number") and the article's firts page.
There are two more bits of information that are usually required:
At least one author's name
And there is one more thing that is more then welcome, but not necessary:
- Article's title.
Every volume is published throughout only one calendar year (but multiple volumes can be published in the same year). The year of the publication of the article is the year of the volume in which it appeared. No matter if it appeared as "e-first" one year before, or if it appeared on arXiv 2 years before, or whatever.
However, if you cite a paper that has appeared as an e-first on the journal's webpage, but has no volume assigned yet, you cite it as "To appear in Journal of Beer Drinking, 2013", even if it's obvious now that it won't be in a volume in year 2013 (since it's already 2014 and the previous volumes are closed). In this case, it's necessary to provide more information to make the citation unique: advised is to provide the article title, which is usually unique in a reasonable time scope.
You should cite the final article where possible. If the article is online but not yet in print, you should cite the DOI which is a permanent resource and will be updated by the publisher when the print volume is announced, so anyone linking to the online version will get the published version. I have noticed some journals taking up to 9 months to assign a print volume to an epub so the DOI is the only sure way to ensure that epub and print citations match up in the end.