Adding Early Access content to Journal Citation Reports: choosing a prospective model

Continuing our discussion of Early Access (EA) content and its planned appearance in the Journal Citation Reports (JCR)™, here we explain why we have chosen a phased, prospective approach to introduce EA content in 2021.

For the past three years, Clarivate has been expanding the number of publishers and journals that have their early access (EA) content indexed in the Web of Science™[i]. As of the end of 2020, EA content from more than 6,000 journals was included, some reaching back to materials published in an EA format in 2017. As we continued to accumulate content, we began to investigate how best to use EA content in the JCR.

Using a dataset of almost 5.3 million source items (517,000 indexed as EA) and 20.6 million citations (nearly 3.6 million either referenced in or linked to EA items), we modeled the outcomes of various methods of including EA content as part of journal performance. Two models are discussed in more detail in our discussion paper: a retroactive model and a prospective model.

Two distinct dates are associated with EA content – an ‘EA date’ marking the first availability of the Version of Record, and a ‘publication date’ tied to a volume-issue-page assignment. Items where the EA date and the publication date fall in different calendar years present a challenge to the calculation of JCR metrics: should these articles be considered in the count of items published in their EA year, or the count of items published in their publication year?


Evaluating retroactive or prospective approaches

The retroactive model would apply EA date backward, onto all content we received as EA from 2017 onward. This would affect the 2020 Journal Impact Factor™ (JIF™) denominator for journals that were providing EA content for indexing in 2018 or 2019, potentially decreasing their JIF value and rank in category as a result of their early participation in our pilot Early Access project. There would be no effect on the JIF denominator for journals that were not providing EA content prior to 2020.

The prospective model would set 2020 as the first year for which EA content would be considered according to its EA date rather than its publication date and would continue to incorporate new content using the EA date. In contrast to the retroactive model, content that was indexed as EA before 2020 would not change the year it is counted in the JCR from its publication date to its EA date.

In both models, content published as EA in 2020 would contribute cited references to the 2020 JIF numerator, even if the item does not have a volume-year assignment until 2021 or later. This expands the number of items and citations contributing to the 2020 JIF numerator of most journals in the JCR regardless of whether they themselves publish EA content that is already indexed in the Web of Science.


Choosing a forward-looking model

We have chosen to implement the prospective model. The retroactive model would create two populations of journals that are differentially affected – based only on when Clarivate began accepting their EA content, not on any change in the citation or publication dynamics of the journal itself.

Imposing a counting disadvantage on a subset of journals while providing a citation benefit to all is not an acceptable option. The prospective model will allow us to continue expanding the number of journals included in EA indexing as we move through 2021 and beyond. Using the EA date for all new content on-going will allow the JCR data to capture the rapid incorporation of published content and the inclusion of more current citations.

For a full overview you can download our discussion paper here.


[i] We define EA content as the public availability of a Version of Record in advance of the assignment of that article to a volume, issue and page. See: