Source-linked AI summary
Self-Selected or Mandated, Open Access Increases Citation Impact for Higher Quality Research
Yassine Gargouri, Chawki Hajjem, Vincent Lariviere, Yves Gingras, Les Carr, Tim Brody, Stevan Harnad
TL;DR
The paper asks whether the OA citation advantage is caused by authors selectively archiving higher-quality work rather than by accessibility itself. It compares self-selected with mandated self-archiving using within-journal citation analyses and regression controls, finding that the advantage persists equally under mandates and is independent of other citation correlates, especially benefiting highly citeable articles.
Problem
The paper tests whether the OA Advantage is causal or instead reflects authors selectively making higher-quality, more citeable articles Open Access.
Method
The study compares self-selected and mandated OA using within-journal mean log citation ratios, matched controls, and logistic regression controlling for major citation correlates.
Results
The OA Advantage remains significant and independent of article age, journal impact factor, authors, references, pages, field, article type, and country, and is as large under mandates as self-selection.
Takeaways & Limitations
The advantage is greatest for highly citeable articles because OA frees users to select and cite accessible work beyond subscription constraints.
Takeaways & Limitations
The possible small additional mandate advantage requires replication because it may reflect chance or sampling error.
Abstract
from arXiv · showhide
Articles whose authors make them Open Access (OA) by self-archiving them online are cited significantly more than articles accessible only to subscribers. Some have suggested that this "OA Advantage" may not be causal but just a self-selection bias, because authors preferentially make higher-quality articles OA. To test this we compared self-selective self-archiving with mandatory self-archiving for a sample of 27,197 articles published 2002-2006 in 1,984 journals. The OA Advantage proved just as high for both. Logistic regression showed that the advantage is independent of other correlates of citations (article age; journal impact factor; number of co-authors, references or pages; field; article type; or country) and greatest for the most highly cited articles. The OA Advantage is real, independent and causal, but skewed. Its size is indeed correlated with quality, just as citations themselves are (the top 20% of articles receive about 80% of all citations). The advantage is greater for the more citeable articles, not because of a quality bias from authors self-selecting what to make OA, but because of a quality advantage, from users self-selecting what to use and cite, freed by OA from the constraints of selective accessibility to subscribers only.
Introduction
Limited subscription access prevents researchers from reaching all potential users, reducing articles’ possible usage and citation impact. Open Access is presented as a way to increase research impact and progress.
- 25,000 journals and refereed conference proceedings publish about 2.5 million articles annually across disciplines, languages, and nations.
- No university or research institution can afford subscriptions to all or most journals its researchers may need.
- Articles whose authors self-archive final drafts average twice as many citations as same-journal, same-year non-OA articles.
- Research impact reflects how much work is used, applied, and built upon by other scientists and scholars.
- OA is therefore framed as increasing research impact and progress, beyond public access rights or general knowledge dissemination.
Self-archiving mandates
Self-archiving recommendations and support have not raised the spontaneous deposit rate substantially above 15%, while mandates were adopted to increase compliance. The study examines whether the citation advantage survives when OA is mandated rather than self-selected.
- Only about 15% of 2.5 million articles published annually worldwide are self-archived by their authors.
- Institutional repositories, requests, recommendations, assistance, rewards, and incentives have not appreciably raised self-archiving above the 15% baseline.
- In surveys, 95% of researchers reported they would self-archive if required by their institutions or funders.
- Southampton’s ECS adopted the first official self-archiving mandate in 2002, followed by mandates from institutions and funders worldwide.
- The study tests whether the OA Advantage reflects author self-selection by comparing self-selected and mandated OA across four institutions with longstanding mandates.
Method
The study compares OA and non-OA articles within the same journals and years, contrasting self-selected with mandated deposits. It uses matched controls, repository and citation data, and mean log citation ratios to reduce comparability and distributional problems.
- The study compares citation counts within the same journal and year for self-selected OA, mandated OA, and corresponding non-OA articles.
- The sample covers articles published from 2002 to 2006, with data collected from four institutional repositories and the Thomson-Reuters citation database.
- For each mandated article, controls were drawn from articles in the same journal, volume, and year, then narrowed to 10 semantically closest matches.
- Verified annual deposits estimated mandate compliance at about 60%, while same-journal/year controls confirmed a spontaneous self-selected baseline of about 15%.
- Method: The analysis compares OA ratios using mean log citation ratios, including O/Ø, OM/OS, OS/ØS, and OM/ØM.
- Method: Logarithms normalize citation data, while within-journal and keyword matching reduce effects from citation-count and journal-average differences.
Results
OA articles were cited significantly more than non-OA articles, and the advantage did not diminish when OA was mandated. The individual-institution results largely matched the combined four-institution pattern.
- OA articles were cited significantly more than non-OA articles, confirming the OA Advantage across the four institutions.
- The pattern for individual institutions was largely the same as the average across all four institutions.
- The mandated advantage was if anything greater, although the smallest effect might reflect chance or sampling error.
- Mean log citation differences were significantly greater than 0 for every OA-versus-non-OA comparison, whether OA was self-selected or mandated.
- The mandated OA Advantage showed no detectable reduction relative to self-selected OA despite compliance rates of 60% versus 15%.
Logistic regression
Logistic regressions evaluated whether OA independently predicts citation ranges while controlling article, journal, field, country, and institutional variables. OA remained significant across citation ranges and JIF intervals, with the largest advantage among highly cited articles and an age interaction in the overall analysis.
- OA independently predicts higher citation counts across every citation range, with the largest advantage for articles receiving 20+ citations.The comparison contrasts 1-5 citations with 20+ citations after controlling other citation correlates.
- The overall Age*OA interaction was significant at both the lowest and highest citation comparisons, and the OA advantage became larger for older articles.The interaction represents a non-additive relation between article age and OA.
- 25% of the sample occupied the highest Journal Impact Factor quartile, which was subdivided into two octiles to preserve sufficient sample sizes.The top quartile spans JIF 1.782 to 29.957.
- In low-JIF journals, the OA advantage increased across higher citation ranges, while OA itself was not significant in the lowest citation range.Age, references, authors, OA, and mandatedness were positively correlated with citations in these journals.
- OA was associated with a significant citation advantage in all Journal Impact Factor intervals and in the sample overall.Within separate JIF regressions, OA and Age remained significant even when the Age*OA interaction disappeared.
Discussion
The OA citation advantage remains independent of author self-selection and other citation correlates, while scaling with article quality and citeability. Its components differ in persistence as OA coverage approaches universality.
- Discussion: OA produces an independent positive citation increase after controlling for article age, journal impact factor, authors, pages, references, article type, field, country, and institution.The result is therefore not explained by imbalance in these citation correlates.
- Discussion: Mandated OA has about the same citation advantage as self-selected OA, despite compliance rates of about 60% versus 15%.This weakens the explanation that authors selectively archive only higher-quality, more citeable work.
- Discussion: The study cannot fully separate early-access effects from postpublication OA effects because reliable deposit-date data are not yet available for a sufficiently large sample.The mandates covered refereed postprints, not prerefereeing preprints.
- Discussion: The OA advantage is greater for higher-impact journals and for papers in higher citation ranges, matching the skewness of citations.In a sample of 708,219 articles, 20% received 80% of citations and 29% received 90%.
- Discussion: OA appears to increase the use and citation of already more usable papers by removing subscription-based access constraints, rather than by making unusable papers citeable.The paper characterizes this as a quality advantage rather than a quality bias.
- Discussion: A possible additional mandate advantage is treated cautiously because the small effect may reflect chance or sampling error until replicated.This proposed effect is distinct from the main OA advantage.
- Discussion: Some OA effects will shrink or disappear as OA rises to 100%, although the quality advantage from broader accessibility is expected to remain.Competitive advantages over non-OA and self-selection bias disappear when all articles are OA.
- Discussion: Early access can accelerate and broaden uptake, increasing the eventual citation quota rather than merely allowing citations to arrive sooner.Download advantages may continue even after competitive OA differences disappear.
Summary and Conclusion
The study addresses whether the OA citation advantage is causal or merely a consequence of authors selectively archiving stronger work. Comparing self-selected and mandated self-archiving, it concludes that OA increases accessibility, use, and citation impact, supporting broader mandates and OA impact measurement.
- Summary and Conclusion: The study tests whether the OA Advantage persists under mandatory self-archiving rather than reflecting author self-selection.The comparison concerns refereed postprints across fields where prepublication self-archiving is not generally dominant.
- Summary and Conclusion: The OA Advantage is real, independent, and causal, with larger effects for more citeable articles because OA increases their accessibility.The authors distinguish this quality advantage from a quality bias caused by selective author archiving.
- Summary and Conclusion: When OA removes accessibility constraints, users can selectively access, use, and cite articles they judge most relevant and high quality.
- Summary and Conclusion: Because only about 15% of articles are spontaneously self-archived, institutions and funders need to mandate self-archiving to reach global 100% OA.The stated goal is to let research achieve its full impact potential without subscription-based access limits.
- Summary and Conclusion: OA metrics are being developed to monitor, analyze, evaluate, credit, and reward research productivity and progress.
Appendix 1
Appendix 1 presents four figures comparing the OA Impact Advantage for self-selected and mandatory OA across Southampton ECS, Queensland UT, Minho, and CERN.
- Figure 12 compares the OA Impact Advantage for self-selected versus mandatory OA for Southampton ECS.
- Figure 13 compares the OA Impact Advantage for self-selected versus mandatory OA for Queensland UT.
- Figure 14 compares the OA Impact Advantage for self-selected versus mandatory OA for Minho.
- Figure 15 compares the OA Impact Advantage for self-selected versus mandatory OA for CERN.
A. JIF1 (JIF < 0,633)
The supplied passages identify logistic-regression tables reporting Exp(ß) values across GIF1, GIF2, GIF3, GIF4, and JIF5.
- Exp(ß) values are reported for logistic regressions for GIF1.
- Exp(ß) values are reported for logistic regressions for GIF2.
- Exp(ß) values are reported for logistic regressions for GIF3.
- Exp(ß) values are reported for logistic regressions for GIF4.
- Exp(ß) values are reported for logistic regressions for JIF5.