Preprints Not Preferred

A recent survey by academic publisher Springer Nature suggests academics strongly prefer to read and cite final versions of journal articles over earlier drafts.

Working with academic networking site ResearchGate, Springer Nature set out to explore how much article versions matter to researchers and whether the final published version of an article, known as the version of record, is perceived to be significantly more valuable than a preprint or an accepted manuscript.

A preprint is a version of a journal article that is shared publicly by the authors before formal peer review. The accepted manuscript is the version of a research article accepted after peer review but before any copy editing or formatting.

Nearly 1,400 ResearchGate users responded to the survey in early 2020. A majority of survey respondents said when given the choice between an earlier version of a journal article and the final published version of record, they would choose the final version, viewing it as the most credible and authoritative source. When citing an article in their own work, 83 percent of respondents said they preferred to use the version of record over earlier versions.

In the survey, nine in 10 researchers said that if they didn’t have a subscription to a journal article, they would take at least one step to try and get access to the article, such as emailing the author to request that they send a copy.

Of those nine in 10 respondents, 35 percent said they would only look for the version of record, while 6 percent said they would only look for a preprint. This finding highlights how much weight researchers place on the version of record, said Mithu Lucraft, marketing director of outreach and open research at Springer Nature.

Analyzing the survey results by geography, discipline and career stage, Springer Nature found little variation in reading and citation behavior. The biggest difference seen was in the field of computer science, where users were most likely to use any version of an article for both reading and citation, said Lucraft.

Springer Nature published the research as part of an ongoing syndication partnership with ResearchGate.

Under this deal, version-of-record Springer Nature articles are uploaded to author profiles on ResearchGate. Users affiliated with institutions that subscribe to the relevant Springer Nature journal can access and download these version-of-record articles as PDFs. ResearchGate users without institutional subscriptions can access the documents as a read-only file.

In late 2020, Springer Nature published an evaluation of the pilot phase of its syndication partnership with ResearchGate.

Much like in the recently published survey, Springer Nature’s 2020 analysis found that ResearchGate users strongly preferred to read and cite version-of-record articles when given the choice -- but the value of this finding was limited, said Lucraft.

“The presentation of the VOR as the default on ResearchGate creates a bias towards the VOR, and it is therefore difficult to compare like-for-like usage across formats,” Lucraft said. “This is why we felt the survey, where researchers expressed preference for the VOR and their reasons for this preference, was so important to do.”

It is not particularly surprising that academics say they would choose version-of-record articles over preprint versions of the same article, said Jessica Polka, executive director of ASAPbio, a group that advocates for the open publication of STEM research.

In a recent analysis of articles published on preprint servers bioRxiv and medRxiv, Polka and her colleagues found relatively few differences between preprint articles and their final published counterparts.

“We found that most preprints don’t undergo discrete changes to their conclusions or the data presented in figures and tables by the time they’re published,” Polka said. “However, in a small fraction of papers, the conclusions do change in a significant way.”

Version-of-record articles are a known quantity in academic circles, Polka said. In some disciplines, the concept of preprinting is relatively new, and it makes sense that users would default to what is familiar to them, she said.

It's often easier for users to find version-of-record journal articles than earlier versions, Polka said.

"Preprint servers often display a prominent link to the published version of an article," Polka said. "It's unfortunately rare for journals to offer reciprocal links."

There is value in making preprints available alongside versions of record, Polka said.

“There are good reasons to cite preprints even when a VOR is available. Information can be removed from a preprint due to journal space restrictions, for the sake of a more coherent narrative or splitting the work into a second study,” Polka said.

For Springer Nature, the survey results support its view that making version-of-record research articles accessible to the public at the point of publication in a journal -- a system known as gold open access -- is the best way forward to increase access to scholarly articles.

“Publishers have long maintained that the VOR is important: as the version that we make available immediately via gold OA, it plays a critical role in the transition to open science,” Lucraft said.

Springer Nature began its partnership with ResearchGate with the aim of helping readers “gain access to the most up-to-date version of an author’s work,” Lucraft said.

When the publisher announced the deal in March 2019, there was some surprise among industry experts, who questioned how Springer Nature would benefit from the arrangement.

ResearchGate previously found itself in hot water with some academic publishers for uploading copies of research articles without publisher permission.

While staff at Springer Nature have worked over the past two years to increase their collaboration with ResearchGate, staff at several other major academic publishers have actively tried to distance their companies from the site and diminish its role in the research information landscape -- an interesting division in approach.

In 2018, publishers including Elsevier and the American Chemical Society formed a group called the Coalition for Responsible Sharing. This group sent thousands of take-down notices to ResearchGate demanding the site remove unauthorized copies of journal articles. Elsevier and ACS have also filed copyright infringement lawsuits against ResearchGate, legal battles that are still working their way through American and German courts.

Springer Nature, on the other hand, is so pleased with its deal with ResearchGate that it declared late last year it planned to continue the arrangement indefinitely.

In May, publisher Wiley also announced plans to work with ResearchGate to “support principled sharing of content,” according to a news release at the time.

In an article published by The Scholarly Kitchen in September, authors Lisa Janicke Hinchliffe and Roger Schonfeld wrote that “we should anticipate future syndication partnerships” between ResearchGate and other publishers. Hinchliffe is a professor and coordinator for information literacy services and instruction in the University Library at the University of Illinois at Urbana-Champaign. Schonfeld is the director of Ithaka S+R's libraries, scholarly communication and museums program.

Given ongoing legal battles, it is unlikely that ResearchGate could ever become a one-stop shop with access to version-of-record articles from all the major publishers, said Will Schweitzer, chief product and customer success officer at Silverchair, a company that supports content management at university presses and scholarly publishers.

Syndication partnerships are common in publishing and have been around forever, Schweitzer said. But some publishers have lingering concerns about “where the full text should reside and who will have the primary relationship with the reader.”

A publisher like Springer Nature has several reasons to partner with ResearchGate.

“They want to provide services to authors and potential authors,” Schweitzer said. “Making content more discoverable is really important to build that relationship with the end user.”

Another key advantage of partnering with ResearchGate is access to more data, which Springer Nature can take to its university subscribers in order to justify renewing institutional subscriptions, said Schweitzer.

“A download is a download as long as you can count it,” said Schweitzer.

Although ResearchGate currently has just two syndication partnerships, the company is eager to bring in other publishers.

"We are learning a lot from the pilots and are in the process of building and rolling out a number of products based on these learnings," said Sven Fund, general manager of publisher solutions at ResearchGate. He added there is a "strong interest" from some publishers to begin similar syndication deals.

For now, there is no money exchanging hands in these agreements, said Fund.

"The incentive for publishers is usage of their content," Fund said. "We will add more advanced features with a publisher benefit in the future."