“We always post a preprint at the same time as we submit to a journal for publication. It is a way to get the results out quickly, before the competition. The peer review process can take several months”, says Torsten Åkesson, professor of particle physics at Lund University.
A preprint is a scientific text that has not yet undergone peer review for publication in a scientific journal, for example. However, the definition may vary between disciplines. The concept is not new – there is a long history of researchers sharing article manuscripts among themselves in various ways – but this has become more widespread in recent years.
In physics, the phenomenon was raised as early as 1991. In order to facilitate the discussion of article manuscripts, the particle physicist Paul Ginsparg created an archive for preprint articles, known today as arXiv. The popularity of the server grew rapidly and now also accepts articles in other fields within physics – mathematics, statistics and computer science, among others.
Uploading is free and the text is freely available. Before the text is published, certain checks are performed, such as ensuring that the content is scientific research, but it has not been peer reviewed.
Today, arXiv contains over 1.3 million articles and is run by Cornell University in New York. Preprints are quoted frequently, and Google Scholar ranks one of the sections of arXiv as number 21 of all scientific journals. Most of the articles uploaded are later published in a journal.
Gets out faster
Anders Eklund, associate professor of medical informatics at Linköping University, says that as the research gets out into the open faster, preprints can move the research frontier forward faster. In his research, he analysed various pieces of software used in the study of brain activity. He concluded that several pieces of software were counting incorrectly and published the results as a preprint before they were submitted to a journal.
“They spread quickly, and there were discussions in social media and via mailing lists. We had conducted the analyses with data available online, and before the article was published in a journal, one of the software manufacturers had verified our results and changed their method on the basis of them”, says Anders Eklund.
Posting preprints is now a natural step for Anders Eklund. In addition to speed, he sees preprints as a possible free form of open access.
The researcher can update the original preprint article when the manuscript is accepted by a journal, thus making the reviewed article freely available. However, such an update is not allowed by all journals.
Anders Eklund says that articles that have undergone peer review are more reliable than preprint. Nevertheless, he believes that preprints are generally quite reliable because the researchers put their name on the manuscript.
“If there are obvious shortcomings, scientists may lose their reputation. If those who publish are unknown, it is harder to trust the results.”
Torsten Åkesson says that when it comes to particle physics research, preprints can sometimes be as reliable as the peer reviewed articles.
“Thanks to extensive internal ex-ante verification, preprints from the major international experimental collaborations in particle physics have the same reliability as the peer reviewed, published articles. On the other hand, experiments with few participating researchers and theoretical works, without the corresponding ex-ante verification organisation, can sometimes go through extensive changes during peer review.”
Has become more common in the life sciences
In biology and medicine, the use of preprints is not as widespread as it is in physics. However, the phenomenon got a hefty boost in 2013, when the server bioRxiv, for preprints in biology and life sciences was started by the Cold Spring Harbor Laboratory (CSHL).
The database has grown rapidly and today contains nearly 20 000 preprints. Recently, a collaboration was announced between the Cold Spring Harbor Laboratory and the open access publisher Public Library of Science (PLOS) which means that article manuscripts sent in to PLOS journals can be automatically posted on bioRxiv.
Jessica Polka, director of the research-driven initiative ASAPbio, welcomes this development. She sees great benefits with the rapid communication preprints can offer, and wants to increase their use within the life sciences.
“Researchers are accustomed to exchanging information with each other before peer review, at meetings and poster presentations, for example. Preprints are an extension of that communication, but allow researchers all over the world to study the information and provide feedback.”
She notes that preprints are also an opportunity for researchers early in their careers to show what they have done, and can make it easier to reach out with things that are harder to publish, such as negative results.
There are risks associated with preprints
However, without peer review, there is the risk of distributing research that does not measure up or of reaching conclusions that are not supported by the data. Jessica Polka believes that some degree of scepticism is called for.
“A preprint is not the same as a peer reviewed article. But the fact that it is a preprint does not necessarily mean that its quality is inferior. Peer review is not perfect either. We have to be a little sceptical about what we read”, she says.
Another challenge concerns the public and media attention to preprints. As a rule, preprint servers mark the item as a preprint that is not peer reviewed. Jessica Polka says that readers must be told what this means.
Preprints with erroneous data could have serious consequences, particularly with regard to some medical research. The server bioRxiv checks before uploading that the material does not pose a risk to health, and currently only allows certain clinical research articles. However, MedaRxiv, a preprint server for results from clinical research, is being developed at the Yale School of Medicine.
Some journals are critical
Nowadays most journals allow a preprint to be published before the manuscript is submitted for potential publication. However, Sara Hägg, associate professor in molecular epidemiology at Karolinska Institutet, points out that not all journals allow this.
“Some journals are critical of the fact that the preprint is available, because they say the novelty value has disappeared. You have to find this out in advance”, she says.
Five years ago, Sara Hägg had no preprints, but today she often posts on bioRxiv in conjunction with the manuscript being sent to a journal.
“Preprints provide the opportunity to refer to your own work that has not yet been published. It is also a way of showing other research groups as quickly as possible what we have done.”
In recent years, the established databases for preprints, or working papers, as they are called in some areas, have been joined by others. Recently, for example, the servers engrXiv, SocArXiv and PsyArXiv were launched for researchers in engineering, social sciences and psychology. Some research funding bodies are now also allowing preprints in their applications.
Still uncommon in the humanities
Nonetheless, says Jutta Haider, associate professor and lecturer in library and information science at Lund University, preprints are still uncommon in the humanities.
“In humanities subjects, it is often the actual text that is the result. The text, its wording and the argumentation has a different status than in many other disciplines. You don’t want to share it that way”, she says.
She believes that the fact that the use of preprints differs between different fields of research can be explained by varying publishing traditions. In her own field, she sees no need for a preprint server. Her assessment is that the function preprints can have – to facilitate communication between researchers – is available in other ways, for example through mailing lists, conferences and institutional seminars.
Jutta Haider has some articles marked ‘preprint’ in the university’s own research database. However, these are reviewed versions that have been accepted for publication. She cannot imagine posting the text as it is submitted to the journal.
“Absolutely not! The comments from peer review, and the changes in the text they lead to, are part of the research. The article is improved by them, often there is no value in publishing what was there before”, she says.