When is a paper published?


When a paper is published could seem obvious but this is not a trivial question.

For some time now, a research article can display several different dates that can prove confusing when trying to work out when a scholarly publication is actually released. In the hardcopy print era, before the Web shook up the academic publishing system, the publication date was always associated with the issued date, the moment at which an article was included in an issue and publicly distributed. This was true because serial publications had to publish articles in groups each time for economic reasons. As the print journals released their articles in periodical issues (biannual, quarterly, monthly, etc.), it was easy to determine when an article was published. The issue indicated the moment. Now, that information is missing on the websites of some academic journals, and we have to estimate the date according to the volume and issue number (e.g. Begell house, ATS Journals).

With the advent of online publishing and the adaptation of print journals to the digital environment, many of the traditional conventions are not necessary. Among them, the grouping of articles in regular issues is technically unnecessary because each paper can be released individually. Although most journals still use the format of volumes and issues, many of the newer electronic journals are giving up this practice (e.g. F1000 Research, eLife, PLOS ONE [which still uses volumes and issues as a convention for citation but not as a way to organize the publication]). This  is causing uncertainty with regard to the publication date because it is common to establish, on one hand, the online publication date and, on the other hand, the date when an article is finally assigned to an issue. This is causing confusion about when an article is published and which date is the actual one.

A possible solution might be to ask Crossref (a place where most of the academic publishers deposit their metadata) when a bibliographic record is initially created. However, if we have a glance at the Crossref API, we can find up to eight dates associated with a publication: Indexed date, posted date, updated date, accepted date, deposited date, created date, published-online date, and issued date. This gets worse when each publisher interprets the meaning of each field differently. For instance, Febs Press (Wiley group) assigns the same date to published-online and issued date, whereas Spinger considers created date and published-online date the same thing, but different to issue date. This multiplicity of dates and meanings does not do anything but increase the confusion about the publication date.

Why is it so important to know exactly when a publication is published?

First, classic bibliometrics was focused almost exclusively on citation impact. This metric has an important delay because it takes a long time from when a paper is published till when it might start to be cited; papers are rarely cited within a year. In this sense, accuracy is not so important because we measure citations by years, not months or days (e.g. Journal Impact Factor). However, the appearance of other metrics (e.g. social media mentions, document views, bookmarking saves) that occur in short time spans, is demanding greater precision. The analysis of these new impact indicators has demonstrated that they happen in the first months, days, or hours in the lifespan of a publication. This immediacy requires the specification of the moment in which an article is made publicly available to understand how and when the social, academic and usage impact is generated. Therefore, current bibliometric studies demand a more precise definition of when a paper is published.

A second reason is related to the research evaluation. The delay between the moment in which a paper is accessible on the publisher’s platform and the moment when it is finally issued can be very long. It is common to see that a paper is accessible in one year but is not printed till the next. This detail could have consequences for the evaluation of research careers when time windows are set up. Important contributions that would improve the assessment of researchers could be excluded because they are published online but not formally assigned to an issue. Many evaluation agencies only consider the final issue date of a publication, in spite of the fact that many papers are accepted and accessible on the Web before that.

At a macro level, when statistics about the aggregate production of organizations and countries are created, the consolidation of those figures would require several years because online papers counted in one year would have to be moved to the next year when they are finally issued. It is crucial to know the precise publication date in order to correctly value the performance of researchers in a time period as well as to obtain a reliable picture about the research effort of organizations and countries.

The third reason is related to the previous point and concerns the difficulties when scientific information systems such as citation indexes or academic search engines come up against different publication dates. Two records of the same publication with different publication dates may be wrongly considered distinct, increasing the risk of duplication. In addition, multiple dates require a great effort of updating because bibliographic databases have to be alert to when the definitive publication date occurs. All these continuous verifications consume much time and many economic resources in the maintenance of scholarly information systems.

What can we do?

As a first step, a clear definition about the meaning of ‘publication date’ should be established. Publishers, through Crossref, must indicate the precise moment in which the paper is definitively released. In my opinion, that date should be the moment when the paper is made publicly accessible on the Web or another medium. That is the time when articles come to life, when they start to be read, mentioned, saved, and cited. Therefore, the online or web date should be considered the principal date for bibliometric studies, research evaluation and information systems. In 2021, Clarivate stated that only the web publishing date (Early Access) would be used for computing the Journal Impact Factor.

The next step would be to define international standards that allow us to respond easily, quickly, and accurately whenever a paper’s publication date is sought.

Author Bio: José Luis Ortega is Tenure Scientist in the Institute for Advanced Social Studies (IESA) of the Spanish National Research Council (CSIC).