Loading evaluation data…
© 2026 The Authors. Published by MIT Press. This is an open access article available under a Creative Commons licence. The published version can be accessed at the following link on the publisher’s website: https://doi.org/10.1162/QSS.a.470
there is no universal OAA across all disciplines or impact indicators
directly identified by consistent heterogeneity across the discipline-by-indicator comparisons
The OAA for citations tends to be lower for more recent papers
visible cohort trend, though unequal citation-accrual windows and disciplinary exceptions limit interpretation
The evidence is consistent with OA accelerating research impact in the Medical & Health Sciences, Life Sciences and the Humanities
unadjusted cross-sectional comparisons cannot identify OA-driven acceleration
OA publications are often associated with higher impacts than non-OA research
broad descriptive ratios support association, but selection and confounding remain unresolved
Derived from the full evaluation — not a separate score.
Strengths
The analysis combines 38.7 million DOI-bearing articles, eleven publication years, six disciplines, and seven indicators, allowing heterogeneity to be examined rather than assuming a universal OA advantage.
↳ Objectives, p. 10; Methods, pp. 10–13
For each indicator, the paper separately analyzes the proportion of articles receiving attention and log-transformed mean attention, distinguishing reach from intensity.
↳ Methods, Analysis, pp. 12–13
The paper cites positive, subset-specific, and null OA findings and compares its results with six competing explanations, including self-selection, early access, and network effects.
↳ Introduction, pp. 7–10; Discussion, pp. 38–40
Limitations
The OA/non-OA comparisons use raw ratios stratified only by broad discipline and year, without adjustment or matching for journal prestige, article quality, author characteristics, funding, geography, or OA route.
↳ Methods, pp. 10–13; Discussion, pp. 38–40
The Abstract and Conclusions discuss OA accelerating, strengthening, or boosting impact, but the cross-sectional snapshot and unadjusted comparisons identify associations rather than those mechanisms.
↳ Abstract; Conclusions, pp. 40–41
The initial Methods description specifies 2011–20, while the objectives, dataset description, figures, and results consistently use 2011–21. This leaves the exact cohort definition needing correction.
↳ Methods, p. 10; Data, pp. 11–12; Results, p. 13
The integrated analysis across disciplines, years, and indicators supports the paper's central descriptive conclusion that no single OA advantage applies universally. Methods appropriately distinguish indicator prevalence from mean intensity and address skew through log-plus-one transformations. The main score constraint is that the OA/non-OA ratios are not adjusted for self-selection, journal prestige, article quality, funding, geography, or OA route, even though the Discussion uses them to compare causal explanations. The paper's extensive qualifications and limitations prevent those interpretive passages from overwhelming the descriptive contribution, but they do not make the mechanisms identifiable.
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Sound3.2
Confidence highThe paper meaningfully extends prior work by integrating 38.7 million articles, eleven publication years, six disciplines, and seven indicators in one comparative analysis. The advance is primarily broader descriptive mapping rather than resolution of causal uncertainty.
“the first complete examination of both the OACA and the OAAA in a very large corpus”
The transformations, cohort definitions, and separate analyses of proportions and means are appropriate for descriptive mapping. No matching, adjustment, sensitivity analysis, or other strategy addresses self-selection, journal prestige, article quality, or related confounding, which binds the score at 2.5.
The research questions, indicator-specific results, and RQ-by-RQ discussion create a followable argument, and most mechanistic interpretations are qualified. Numerical density, acronym rotation, and causal-adjacent phrases such as OA “accelerating” impact reduce precision at the claim level.
The paper engages mixed prior findings and identifies limitations involving DOI coverage, OA-route aggregation, unequal accrual time, and status switching. Self-selection is discussed, although the implications of uncontrolled confounding are not fully integrated into the mechanism comparisons.
“OA status is therefore not the only confounding factor contributing to the impact of research”
Lower confidence on Reporting, Positioning — domain match limited.
Caveats4 of 4 checks
The manuscript contains a date-range inconsistency in its definition of the analytic cohort and an informally handled large Wikipedia-citation outlier. These are watch-outs for exact cohort and ratio interpretation rather than evidence that the broad descriptive results are unreliable.
Employment by the company associated with the principal bibliometric and altmetric data sources creates a structural conflict, but it is disclosed. The study provides a persistent repository identifier, and no human-participant ethics declaration is expected for this metadata analysis.
Flags: 2 declared / 5 total
73 of 73 checkable references verified
82 references in manuscript 9 have no canonical index record — counted, but not index-checkable citation details diverge from the cited records for 1 reference 3 references confirmed by manual review
No retraction notice found in Retraction Watch.
Sources: Retraction Watch ✓
Where this paper’s evidence sits on the path from initial observation to real-world use.
The analysis uses full-scale observational data but does not evaluate an OA intervention, mandate, or implementation strategy. Consequently, it provides contextual evidence rather than deployment-ready guidance.
AI-generated, human-governed. Something look off? Contact us to request a review.