{"id":"155d103a-708f-4e81-8d8e-9c93b7bb881f","arxiv_id":"2411.14653","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Open access papers are cited more often by papers from other fields in 13 of 18 natural science fields, with clinical medicine showing a robust interdisciplinary citation advantage.","lead":"Using Web of Science data from 2017, the authors measure whether gold open access papers receive more citations from outside their own field than non-open access papers. They find that open access is associated with more cross-field citations in most natural sciences, notably in clinical medicine, and introduce metrics that split the open access citation advantage into interdisciplinary and within-discipline parts.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The main IOACA is an unadjusted field-level mean comparison; the paper's own Fig. 4 and Tables 2–3 show chemistry and computer science results are driven by a few multidisciplinary gold-OA megajournals, so the causal 'OA increases' claim is not supported.","rationale":"The reader's weakest assumption named residual confounding, and that is correct, but my concern is more specific: the main metric is not actually a JIF-quartile-controlled estimate. The paper's own Fig. 4 and Tables 2–3 show that composition and a few megajournals drive the field-level IOACA in chemistry and computer science. This is the load-bearing link because a positive IOACA is entirely consistent with no OA effect if gold OA papers are concentrated in journals that attract cross-disciplinary citations regardless of openness. The clinical medicine case study uses topic similarity, but that is not an identification strategy: it demonstrates that cited topics are similar, not that access, rather than paper selection, caused the extra citations. Nonetheless, the paper is transparent, the decomposition is clearly defined, and the data construction is reproducible in principle from WoS and DOAJ. As a descriptive bibliometric association, the work is usable; the condition for publication is that the causal language be softened or supported by adjusted estimates. The reader's CONDITIONAL verdict already captures this, so my read does not move the verdict.","tokens_in":12993,"tokens_out":5561,"duration_ms":59984,"concrete_test":"Core fix: recompute all field-level IOACA/WOACA values with a within-JIF-quartile pooling (inverse-variance weighted) or a matched design, and compare to the raw Fig. 1 values with bootstrap confidence intervals. If chemistry/computer science/clinical medicine adjusted IOACAs collapse or flip sign, the causal claim fails; if they survive, the megajournal concern is substantially weakened.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central causal claim that gold OA increases interdisciplinary citations rests on the IOACA metric, but that metric is a raw ratio of means calculated at the field level, with no model-based adjustment. The JIF-quartile breakdown described under 'JIF quartile' is not integrated into a controlled estimate; it is presented descriptively as Fig. 4. The paper itself warns that the raw field-level OACAs 'may have been influenced by the proportion of gold OA journals in higher JIF quartiles within a field' (Results, 'OACAs by field'), and Tables 2–3 make the consequence concrete: in chemistry, three Q2 gold-OA journals (IJMS, Sensors, Molecules) account for 82% of Q2 gold-OA interdisciplinary citations; in computer science, two journals (Journal of Statistical Software and BMC Bioinformatics) account for 92% of Q1 gold-OA interdisciplinary citations. Since gold OA status is assigned at the journal level via DOAJ membership, the OA/non-OA contrast is largely a contrast between journal portfolios and subfields. The field-level IOACA for chemistry and computer science therefore reflects a small set of highly cited multidisciplinary megajournals, not a general effect of OA on the average paper. The clinical medicine result is more plausibly field-wide, but it is still based on an unadjusted aggregate comparison; the topic-similarity case study does not rule out unmeasured differences between papers in gold-OA and subscription journals. Without a within-quartile or matched estimate with uncertainty, the phrase 'OA increases' is not supported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a decomposition of the open access citation advantage (OACA) into an interdisciplinary component (IOACA) and a within-discipline component (WOACA), and applies this decomposition to Web of Science data for 2017 gold-OA and non-OA articles in 18 natural science fields with a six-year citation window. The authors report that in 13 fields both IOACA and WOACA are positive, while in chemistry, computer science, and clinical medicine only IOACA is positive. They then present a case study of clinical medicine using citation-topic cosine similarities to argue that gold OA increases interdisciplinary citations by improving accessibility rather than by publishing topically different papers.","tokens_in":13224,"tokens_out":2794,"duration_ms":29335,"significance":"If the causal interpretation were supported, the paper would make a valuable contribution to the OA citation advantage literature and to policy discussions about knowledge transfer across disciplines. The decomposition is a useful descriptive tool, the dataset is large and well described, and the field-level comparisons are transparent. The authors also deserve credit for explicitly investigating the role of specific journals and for presenting robustness checks by year and by JIF quartile. However, the central claim that 'OA increases interdisciplinary citations' is not supported by the evidence as presented, because the measure is an unadjusted aggregate comparison and the paper's own journal-level analyses show that several field-level results are driven by a small number of multidisciplinary megajournals. The clinical medicine case study is more plausible but still observational.","major_comments":[{"comment":"The central causal claim that gold OA increases citations, including interdisciplinary citations, rests on the IOACA, which is an unadjusted ratio of field-level means. The JIF-quartile stratification is described in the Methods and shown descriptively in Fig. 4, but it is not used to produce controlled estimates; no within-quartile averages, regression adjustment, or matching is performed. The authors themselves state in the Results that the raw OACAs 'may have been influenced by the proportion of gold OA journals in higher JIF quartiles within a field.' Because gold OA status is assigned at the journal level, the OA/non-OA contrast reflects differences between journal portfolios and subfields, not just the effect of open access on otherwise comparable papers. The abstract and conclusion use causal language ('OA increases'), which is not justified by the design. I recommend either reframing the findings as descriptive associations or adding an adjusted analysis, such as computing IOACA within JIF quartiles and pooling, or a regression with JIF quartile and field fixed effects.","section":"Methods, 'JIF quartile'; Results, 'OACAs by field'"},{"comment":"The paper's own journal-level tables show that the field-level IOACA for chemistry and computer science is driven by a very small set of multidisciplinary gold-OA journals. In chemistry, three Q2 journals (International Journal of Molecular Sciences, Sensors, Molecules) account for 82% of Q2 gold-OA interdisciplinary citations; in computer science, two journals account for 92% of Q1 gold-OA interdisciplinary citations. The authors acknowledge this, but they still present the field-level IOACA as evidence that OA fosters interdisciplinary citation in these fields. This is a load-bearing issue: the headline claim that OA increases interdisciplinary citations in chemistry and computer science cannot be distinguished from the claim that a few megajournals with broad scope attract interdisciplinary citations. I suggest a robustness analysis that recomputes the field-level IOACA excluding these journals, or at least a quantitative concentration measure, and a correspondingly cautious interpretation if the results disappear.","section":"Results, 'Distribution of interdisciplinary citations by journal'; Tables 2 and 3"},{"comment":"The clinical medicine case study is the strongest part of the paper because the high IOACA appears across JIF quartiles rather than being concentrated in a few journals. However, the interpretation that 'OA has improved these papers' accessibility' (and therefore caused more interdisciplinary citations) is a causal claim that is not identified from the data. The cosine similarity between OA and non-OA citation-topic distributions shows that the topics of the two groups are similar, but it does not rule out unmeasured confounders such as author prestige, institutional reputation, funding, or the authors' deliberate choice of a broader-audience journal. The comparison remains observational. I recommend softening the causal language and explicitly discussing residual confounding, or, ideally, supplementing the analysis with a quasi-experimental approach such as a natural experiment involving journal flips to OA or a within-journal comparison over time.","section":"Results, 'Case study of clinical medicine'; Table 4"}],"minor_comments":[{"comment":"The paper would benefit from uncertainty quantification. Although the dataset is the total population of eligible WoS papers, the interpretation of differences between fields and between groups would be stronger with confidence intervals, bootstrapped standard errors, or at least a significance test. The current presentation of every difference as meaningful makes it hard to know which findings are stable.","section":"Throughout"},{"comment":"The text says DOAJ data were used to 'identify cited papers definitively categorized as gold OA in 2017.' A DOAJ listing is not necessarily definitive for the publication year 2017, because journals can enter the directory later. Clarify how the temporal match was performed (e.g., whether the DOAJ record used reflects the journal's status in 2017 or at the time of data access).","section":"Methods, 'OA type'"},{"comment":"The titles of Tables 2 and 3 suggest that the listed journals represent the entire distribution, and the 'Total' row sums to 100%, but the text implies these are only the top journals. Please state explicitly whether these tables cover all Q2 gold-OA chemistry journals and all Q1 gold-OA computer science journals, or only a selected subset, and clarify what the total represents.","section":"Results, 'Distribution of interdisciplinary citations by journal'; Tables 2 and 3"},{"comment":"In the text near Table 4, the phrase 'oa_same and no_same' appears to be a typo for 'oa_same and non_same.' Please correct.","section":"Results, 'Case study of clinical medicine'; Table 4"},{"comment":"In several places the paper uses causal phrasing such as 'OA has been shown to foster knowledge transfer' and 'OA effectively enhances interdisciplinary citations.' Given the observational design and the issues raised in the major comments, these statements should be replaced with associational formulations unless additional analyses are added.","section":"Discussion and Conclusion"}],"recommendation":"major_revision","confidential_remarks":"The paper is a good fit for a scientometrics or bibliometrics journal, and the decomposition idea is a nice contribution. The main substantive problem is the gap between the descriptive, unadjusted analyses and the causal language in the abstract and conclusion. I believe this is fixable within the manuscript's scope by reframing the claims and adding robustness checks, so I recommend major revision rather than rejection. I would also encourage the editor to ensure the reviewers look carefully at whether the journal-level results in Tables 2 and 3 undermine the field-level conclusions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe one thing to know: this is a descriptive bibliometric decomposition, not a causal study, despite the causal verb in the abstract. The new measurement is the split of the OA citation advantage into interdisciplinary (IOACA) and within-discipline (WOACA) components, applied to 18 natural science fields using 2017 WoS papers and a 6-year citation window. That decomposition is simple, transparent, and genuinely new relative to prior work like Huang et al. (2024), which used diversity indices rather than citation counts. The field-level results, especially Group 2 fields (chemistry, computer science, clinical medicine) where only the interdisciplinary component is positive, are a real addition to the OA advantage literature. The clinical medicine case study—comparing topic cosine similarities between OA and non-OA papers cited from different field groups—is a thoughtful attempt to argue for an accessibility mechanism, and the finding that topic distributions are similar while citation patterns differ is worth taking seriously.\n\nThe soft spots are the usual ones for observational OA studies, and they are concentrated in the causal interpretation. The IOACA is an unadjusted ratio of means at the field level. The only control is JIF quartile, presented descriptively in Fig. 4, not integrated into a model or matched estimate. There are no confidence intervals or significance tests anywhere in the paper. More concretely, the paper's own Tables 2–3 show that the chemistry and computer science results are driven by a handful of multidisciplinary gold-OA megajournals (IJMS, Sensors, Molecules in chemistry; J. Statistical Software, BMC Bioinformatics in CS). The authors acknowledge this in the discussion, but the abstract still says OA 'increases' interdisciplinary citations in those fields. That is a genuine overstatement. The clinical medicine result is more plausibly field-wide, but it still rests on the same unadjusted aggregate comparison; the topic similarity analysis reduces some confound concerns but does not eliminate them. The paper's own limitation section is honest about ESI coarseness, the natural-science-only scope, and gold-OA-only focus, but it does not flag the missing uncertainty quantification.\n\nBottom line: as a descriptive contribution introducing a new metric and field-specific patterns, this is solid and worth a serious referee. As a causal claim about OA fostering knowledge transfer, it is not supported by the design. A revision that rewords causal claims to associational ones, adds at least per-field CIs or a within-quartile analysis, and reports the megajournal sensitivity prominently would make it publishable in a good scientometrics venue. I'd recommend sending to peer review rather than desk rejecting.","headline":"Useful descriptive decomposition of the OA citation advantage, but the causal framing outruns the unadjusted, megajournal-driven evidence.","tokens_in":13778,"tokens_out":2033,"would_cite":true,"duration_ms":19128,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Gold open access raises cross-field citations in 13 of 18 sciences.","keywords":["open access citation advantage","gold open access","interdisciplinary citation","within-discipline citation","knowledge transfer","citation analysis","bibliometrics"],"falsifier":"The claim would be overturned if the interdisciplinary citation advantage vanished in a comparison that matched gold-OA and non-OA papers on author record, funding, preprint availability, and journal, or in a natural experiment where a journal's move to gold OA changed access while authors and submission pools stayed fixed.","tokens_in":12728,"feed_emoji":"🔓","tokens_out":7603,"duration_ms":67211,"temperature":0.7,"pith_summary":"This paper asks whether open access does more than raise citation counts: does it carry findings into other disciplines? It claims that gold OA increases interdisciplinary citations in 13 of the 18 natural-science fields studied, and that in chemistry, computer science, and clinical medicine the effect is exclusively interdisciplinary. To reach that conclusion, the authors split the standard open access citation advantage into two parts, one counting citations from other fields and one counting citations from the same field. If the claim holds, OA is a concrete policy lever for cross-disciplinary knowledge transfer, not just a visibility booster.","feed_headline":"Gold open access pulls citations across field borders","feed_subtitle":"In 13 of 18 natural-science fields, OA papers draw more citations from other disciplines, not just their own.","key_machinery":"The load-bearing instrument is the decomposition of the OACA into $IOACA$ and $WOACA$, applied to all gold-OA and non-OA articles and reviews published in 2017 in 18 natural-science fields, with citations counted over a six-year window and papers stratified by journal impact factor quartile. The decomposition turns a total citation count into a directional measure of cross-field flow: a positive $IOACA$ with a negative or nil $WOACA$ identifies fields where OA's effect is specifically interdisciplinary. The journal-quartile stratification and citation window are the controls used to dampen the two classic confounds of OA-citation studies, early online availability and selection of high-quality papers into OA.","core_discovery":"The paper's central claim is that gold OA increases the number of citations a paper receives from other fields in 13 of 18 natural-science fields, and that in chemistry, computer science, and clinical medicine this interdisciplinary advantage exists while the within-discipline advantage does not. The authors reach this by decomposing the open access citation advantage (OACA), the percentage by which average citations of gold-OA papers exceed those of comparable non-OA papers, into two parts: the interdisciplinary OACA ($IOACA$), based only on citations from papers outside the field, and the within-discipline OACA ($WOACA$), based only on citations from the same field. They interpret positive $IOACA$ values as evidence that OA fosters knowledge transfer to other fields.","pith_inferences":["If the association is causal, gold OA is one of the few concrete policy levers available to increase the flow of findings into neighboring disciplines, alongside other levers such as data sharing or plain-language summaries.","The same decomposition applied to green OA would clarify whether the cross-field effect comes from accessibility itself or from the publication venue, since green OA separates free availability from journal choice.","For fields with no interdisciplinary advantage, the result suggests that access alone may not change citation flows; field-specific indexing and disciplinary reading habits may dominate.","A sharper test would compare the same journal before and after flipping to gold OA, or compare papers subject to funder OA mandates, to see whether the interdisciplinary share of citations rises when access changes."],"forward_implications":["In 13 of 18 natural-science fields, gold OA papers receive more interdisciplinary citations than non-OA papers, so OA can serve as a channel for cross-disciplinary knowledge transfer.","In chemistry, computer science, and clinical medicine, gold OA shows an interdisciplinary citation advantage without a within-discipline advantage, meaning cross-field readers are the ones gained.","In agricultural science, engineering, environment/ecology, and materials science, gold OA shows no advantage of either type, so the knowledge-transfer effect is not universal.","In clinical medicine, the effect is not explained by a few journals or by OA papers covering different topics; OA and non-OA papers have similar citation topics, and OA papers are cited more from nearby fields, pointing to accessibility as the driver.","At the field level, some advantages are journal-driven: chemistry's Q2 gold journals and computer science's Q1 gold journals account for most of their fields' interdisciplinary citations."],"supporting_citations":[{"why":"Supplies the OACA metric and the multi-field gold OA analysis that this study decomposes.","marker":"Dorta-González et al., 2017"},{"why":"Supplies the OACA formula and the six-year citation window used to control for the early-view effect.","marker":"Sotudeh et al., 2015"},{"why":"Supplies the journal impact factor quartile stratification used to address selection bias.","marker":"Niyazov et al., 2016"},{"why":"Provides the recent multi-measure OA citation advantage analysis whose field-level approach this study follows.","marker":"Basson et al., 2021"},{"why":"Supplies the prior finding that OA outputs receive more diverse citations, which this study makes field-specific.","marker":"Huang et al., 2024"},{"why":"Supplies the two-journal evidence that OA papers receive more interdisciplinary citations, which this study generalizes.","marker":"Young and Brandes, 2020"},{"why":"Documents that clinical medicine receives few citations from other disciplines, the baseline the clinical-medicine case study qualifies.","marker":"Van Noorden, 2015"},{"why":"Documents the lack of consensus on the OA citation advantage, motivating the new decomposition.","marker":"Langham-Putrow et al., 2021"}],"fun_headline_variants":["Gold OA boosts cross-field citations in 13 of 18 fields","Open access drives interdisciplinary citation gains","Gold open access widens citation reach beyond field","OA fosters knowledge spillover: 13 fields gain cross-discipline cites","Interdisciplinary citation advantage from gold OA"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that, after comparing gold-OA with non-OA papers within the same journal-impact-factor quartile, the remaining difference in citations is caused by open access rather than by unmeasured differences such as paper quality, author prestige, funding, or institutional reputation.","fun_headline_variants_meta":{"raw":{"variants":["Gold OA boosts cross-field citations in 13 of 18 fields","Open access drives interdisciplinary citation gains","Gold open access widens citation reach beyond field","OA fosters knowledge spillover: 13 fields gain cross-discipline cites","Interdisciplinary citation advantage from gold OA"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00016,"raw_usage":{"total_tokens":1185,"prompt_tokens":849,"completion_tokens":336,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":465,"completion_tokens_details":{"reasoning_tokens":261}},"tokens_in":465,"tokens_out":336,"duration_ms":3465,"temperature":1.0,"reasoning_tokens":261,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:02:36.292433+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The claim would be overturned if the interdisciplinary citation advantage vanished in a comparison that matched gold-OA and non-OA papers on author record, funding, preprint availability, and journal, or in a natural experiment where a journal's move to gold OA changed access while authors and submission pools stayed fixed.","supporting_citations":[],"review_version":1}