Does Open Access Increase Citations?

Open articles often collect more citations, but the strongest studies disagree on why. Here is what each kind of evidence can and cannot tell you.

Explainer · 9 min read · Directive Publications Editorial Team

The short answer

Open access articles are often cited more than paywalled ones. Whether openness is the reason is still argued over, more than twenty years after the first studies.

That gap between correlation and cause is the whole debate. An open article might collect more citations because more people can read it. It might also collect more because authors chose to make their best work open, because it appeared earlier as a preprint, or because it sits in a field and journal that are cited heavily anyway. Most studies can see the first effect but cannot rule out the others.

For a research office, the practical answer is this: open access reliably widens readership, and some well-designed studies find a modest citation gain on top of that, but no study design supports promising an individual author a fixed percentage boost.

Why the question is so hard to answer

A clean test would take identical papers, make half of them open at random, and count citations later. Almost no study can do that, so most compare articles that became open by choice with articles that did not. Several things get mixed into that comparison.

• Self-selection, or quality bias. Authors may make their strongest papers open, or the most active researchers may be the ones who self-archive. Those papers would be cited more regardless of access.
• Early access. A paper posted as a preprint months before publication has had longer to be found and cited.
• Field and journal. Citation habits vary enormously between disciplines, and open access rates vary with them.
• Route to openness. Green, gold, hybrid and bronze are different things, and they do not behave the same way in citation data.
• Citation database. Each index covers different journals, so the same question can give different answers depending on where citations were counted.

The 2021 systematic review in PLOS ONE by Langham-Putrow, Bakker and Riegelman found that only 40 of its 134 included studies acknowledged the possibility of confounders, and not all of those went on to control for them. Nearly 98% of the studies were judged at high overall risk of bias.

What the large observational studies found

The early studies made the citation advantage famous.

In 2001, Steve Lawrence published an analysis in Nature of roughly 120,000 conference papers in computer science and related disciplines and found that articles freely available online were cited substantially more than those that were not. In 2006, Gunther Eysenbach followed a cohort of articles published in PNAS in 2004, where authors could pay a fee to make their article open. After adjusting for factors such as number of authors, authors' previous publication and citation record, submission track, country and funder, open articles were about twice as likely to be cited in the first 4 to 10 months (odds ratio 2.1). He also found that articles open on the journal site had more impact than articles made available only through self-archiving.

The largest recent analysis is Piwowar and colleagues' 2018 PeerJ study, "The state of OA". In a random sample of Web of Science articles and reviews published between 2009 and 2015, open articles received 18% more citations than expected for their age and discipline, while paywalled articles were cited 10% below the world average.

All three are observational. They show that open articles are cited more. They cannot show that openness caused it, and Eysenbach's design still depended on authors choosing whether to pay for open access.

What the stronger designs found

Researchers who wanted to separate access from author choice have tried four main approaches. Their results do not line up neatly, which is why the debate continues.

The randomised trials are the cleanest test of access alone, because publishers, not authors, decided which articles were open. They found more readers but no extra citations. The mandate and embargo studies also take the choice away from individual authors, and they do find an advantage. Ottaviani's articles were opened through blanket agreements between publishers and the library, in some cases after their prime citation years had passed.

So the designs that take the choice away from authors do not agree with each other. The trials found no citation effect, the mandate and embargo studies found one, and the journal-level panel found a far more modest one than the early observational studies. Where a gain does appear, several of these studies found it concentrated in articles or journals that were already doing well. The trials also covered particular journals and time windows, so they do not settle what happens in every field.

What the systematic review concluded

Langham-Putrow and colleagues pooled the literature up to their search date. Of 134 studies, 64 reported a citation advantage, 37 found none, 32 found one only in some subsets, and one was inconclusive. The three studies judged at low risk of bias split the same three ways. Their conclusion was that the quality and heterogeneity of the studies make generalisation difficult, and they called for reporting guidelines for bibliometric research.

So "most studies find an advantage" is technically true, and it is also the least useful way to summarise this evidence. Counting studies treats a small, high-bias comparison the same as a randomised trial.

Why the route to openness matters

"Open access" in a citation study can mean very different things, and the 2018 Piwowar analysis shows how much that changes the result.

In its Web of Science sample, articles available only as green open access (a copy in a repository) and hybrid articles (paid-for open articles in subscription journals) were cited more than 30% above expectations. Bronze articles, free on the publisher's site without an open licence, were cited 22% above the world average. Gold articles in fully open access journals were cited 17% below the world average, and below paywalled articles. The authors suggested this was likely due partly to the growth of newer and smaller open access journals, some from regions or languages historically less covered by Web of Science, and partly to the growth of mega-journals.

Two lessons follow for institutions. First, a "citation advantage" figure is meaningless without knowing which kind of open access it describes. Second, the route with the highest relative citation rate in that analysis, green, typically involves no article processing charge. If your office supports self-archiving in a repository, that is worth knowing, though it is evidence of association, not proof that depositing will lift a particular paper.

Beyond the count: who reads and who cites

The argument about citation totals can hide what open access is more clearly shown to do.

Both Davis trials found that open articles reached more readers. The 2011 trial concluded that the real beneficiaries may be communities that read research but rarely publish it, such as practitioners. In medicine that includes clinicians, guideline developers, patients and health workers outside well-funded institutions. Citations do not count any of them.

A 2024 Scientometrics study by Huang and colleagues looked at the question differently. Instead of asking whether open outputs were cited more often, it asked whether they were cited by a wider range of people. Using large-scale bibliographic data from 2010 to 2019, it found a robust association between open access and greater diversity of citing institutions, countries, regions and fields, for both highly cited and less-cited outputs. Open access through disciplinary or institutional repositories showed a stronger effect than open access through publisher platforms.

That is still an association, but it is closer to what open access was designed to achieve. For measures of reach that go beyond citations, see our guide to open access metrics: downloads, altmetrics and reach.

How research offices should use this evidence

The evidence supports a clear, defensible position. Turn it into wording your office can reuse.

In policy documents and author guidance, lead with what the evidence supports firmly: wider readership, reuse, funder compliance and access for readers outside subscribing institutions. If you mention citations, say that many studies find open articles are cited more, that studies which remove author choice give mixed results, from no effect to a modest one, and that the size of any effect varies by field and route.

When you meet a headline figure in a vendor report, a publisher pitch or an internal presentation, ask five questions before repeating it:

1. Was openness assigned by authors, or by something outside their control?
2. Which kind of open access was measured: gold, green, hybrid or bronze?
3. Were field, article age and journal accounted for?
4. Which citation database was used, and over what time window?
5. Is the figure an average across all papers, or concentrated in highly cited ones?

In research assessment, keep citation counts out of decisions about whether open access spending was worthwhile. Promotion and hiring panels comparing open and paywalled papers should know that a raw citation difference says little about the quality of the work or the author.

When talking to individual authors, be honest. Making a paper open may help it be found and read, and it may help citations. Nobody can promise a specific increase, and authors who publish open access mainly for a citation boost may be disappointed.

The bottom line

Does open access increase citations? Sometimes, modestly, and more reliably for some routes and some papers than others. The evidence is much stronger that it increases readership and broadens who can use research.

That is a better reason to support open access than a citation promise. It is also one your office can defend in front of a sceptical committee. For the wider picture of models, licences and costs in medicine, see Open Access Publishing in Medicine: The Complete Guide.