TL;DR
Open Access (OA) articles are 64.7% more likely to be cited on Wikipedia than paywalled research, even after controlling for field and academic impact. While only about 23% of global scientific literature is freely available, Wikipedia "over-indexes" on open science, with 44.1% of its cited publications being Open Access.
What Problem Does This Paper Address?
Wikipedia is the world’s largest repository of knowledge, but its reliability depends entirely on its citations. For years, the "Open Access" movement has argued that removing paywalls accelerates the dissemination of science. Intuitively, we expect Wikipedia editors—who are often volunteers without institutional library access—to favor free articles.
However, quantifying this "OA citation advantage" has been difficult. Previous studies often looked at the journal level (e.g., "Is this an OA journal?") rather than the article level. This missed "Green OA" (paywalled articles with free repository copies) and "Hybrid OA" (individual free articles in paid journals). This paper seeks to provide a granular, article-level analysis to determine exactly how much a paywall hinders a discovery from reaching the general public via Wikipedia.
Methods
The researchers employed logistic regression modeling to estimate the probability of an article being cited on Wikipedia.
To ensure a fair comparison, they used stratified sampling to create a balanced dataset. They didn't just look at what was cited; they compared cited articles against a "negative sample" of articles that were not cited but shared similar characteristics (same publication year, same field of study, and similar journal prestige).
The models accounted for several confounding variables:
is_oa: The binary status of whether the paper is free to read.article_age: How long the paper has been out.SJR: The Scimago Journal Rank (a proxy for prestige).concept: The scientific field (e.g., Biology vs. History).times_cited: The article's existing academic impact in the scholarly community.
Datasets
The study integrated three massive data sources:
- Wikipedia Citations (Feb 2024): A dataset of over 45 million citations extracted from English Wikipedia, narrowed down to ~2.2 million journal articles with DOIs.
- OpenAlex: A massive open index of scholarly works used to retrieve article-level metadata, including OA status, specific OA policies (Gold, Green, Bronze, Hybrid), and citation counts.
- Scimago: Used to obtain the SCImago Journal Rank (SJR) to control for the influence of high-prestige journals like Nature or Science.
Key Insights and Ideas
- Accessibility Over Prestige: In a surprising twist, the model showed that as a journal's prestige (SJR) increases, the likelihood of its articles being cited on Wikipedia actually decreases slightly when other factors are held constant. This suggests Wikipedia editors prioritize the ease of verifying a fact over the "brand name" of the journal.
- The "Freshness" Factor: The OA advantage is most pronounced for recent research. Wikipedia editors are highly responsive to new developments, frequently updating pages with OA articles published within the last four years.
- STEM vs. Humanities: The reliance on OA is not uniform. STEM fields (Biology, Physics, Mathematics) show a much stronger dependence on OA sources than the Humanities. For instance, 57% of Biology citations on Wikipedia are OA, compared to only 24% in History.
- The Power of "Green" and "Bronze": The study highlights that "Bronze" OA (free to read on a publisher site but without a formal license) and "Green" OA (self-archived copies) are the most common types of free science on Wikipedia, proving that repository-based sharing is a vital bridge for public knowledge.
Key Results
- The 64.7% Boost: After controlling for an article’s age, field, and existing academic citations, an OA article is 64.7% more likely to be cited on Wikipedia than a paywalled counterpart.
- Higher Adoption Rates: Wikipedia’s reliance on OA (44.1% of publications) significantly outpaces the general availability of OA in databases like Web of Science (23.6%) and OpenAlex (22.6%).
- Impact Correlation: The "OA advantage" is particularly strong for articles that already have high citation counts in the academic world. Essentially, if a paper is both impactful and free, its chances of hitting Wikipedia skyrocket.
Limitations and Future Work
The authors acknowledge several constraints:
- DOI-Only Focus: By focusing on articles with Digital Object Identifiers (DOIs), the study excludes conference papers, books, and older literature that may not have modern identifiers.
- Static Snapshot: The study doesn't account for when a citation was added. Future work could analyze Wikipedia’s edit history to see if an article was cited specifically because it flipped from paywalled to OA.
- The "OABOT" Effect: Wikipedia uses tools like OABOT to automatically find OA versions of paywalled citations. The study doesn't fully decouple how much of the OA prevalence is due to editor choice versus automated "upgrading" of citations.
- Beyond Citations: The researchers suggest that future studies should look at other "Open Science" pillars, such as how open data and open software influence Wikipedia’s reliability.
Summary updated