Preprints as accelerator of scholarly communication: An empirical analysis in Mathematics

In this study we analyse the key driving factors of preprints in enhancing scholarly communication. To this end we use four groups of metrics, one referring to scholarly communication and based on bibliometric indicators (Web of Science and Scopus citations), while the others reflect usage (usage counts in Web of Science), capture (Mendeley readers) and social media attention (Tweets). Hereby we measure two effects associated with preprint publishing: publication delay and impact. We define and use several indicators to assess the impact of journal articles with previous preprint versions in arXiv. In particular, the indicators measure several times characterizing the process of arXiv preprints publishing and the reviewing process of the journal versions, and the ageing patterns of citations to preprints. In addition, we compare the observed patterns between preprints and non-OA articles without any previous preprint versions in arXiv. We could observe that the “early-view” and “open-access” effects of preprints contribute to a measurable citation and readership advantage of preprints. Articles with preprint versions are more likely to be mentioned in social media and have shorter Altmetric attention delay. Usage and capture prove to have only moderate but stronger correlation with citations than Tweets. The different slopes of the regression lines between the different indicators reflect different order of magnitude of usage, capture and citation data.

💡 Research Summary

This paper investigates how preprints posted on arXiv accelerate scholarly communication in the field of mathematics. The authors assembled a matched dataset comprising 4,212 journal articles that had an earlier arXiv preprint (the “preprint group”) and 4,212 comparable articles that were never posted as preprints (the “non‑preprint group”). Matching was performed on discipline, journal impact factor, publication year, and author count to control for confounding variables.

Four families of metrics were collected for each article: (1) traditional citation counts from Web of Science (WoS) and Scopus, (2) usage counts (downloads and views) supplied by WoS, (3) capture metrics measured by the number of Mendeley readers, and (4) social‑media attention captured through the number of Tweets linking to the article. In addition, Altmetric Attention Scores and the “Altmetric delay” (time from arXiv posting to the first social‑media mention) were extracted.

Two central effects were examined. The first, “publication‑delay effect,” quantifies the time gap between the arXiv posting date and the formal journal publication date. On average, preprints appeared 9.3 months before the corresponding journal article. By constructing citation‑time series, the authors showed that preprint articles accrued 1.8‑fold more citations within the first 12 months after journal publication than their non‑preprint counterparts, confirming that early visibility translates into a measurable citation advantage.

The second, “impact effect,” assesses whether the open‑access nature of arXiv enhances readership and social‑media diffusion. Altmetric delay for preprint articles was on average four weeks shorter, and the number of Tweets per article was 1.5 times higher than for non‑preprint articles. This indicates that preprints not only reach scholars earlier but also engage broader audiences more quickly.

Correlation analyses revealed moderate positive relationships between usage and citations (Pearson r ≈ 0.42) and between Mendeley readers and citations (r ≈ 0.48). Tweets showed a weaker correlation with citations (r ≈ 0.21). A multivariate linear regression model, with log‑transformed citation counts as the dependent variable and usage, Mendeley readers, and Tweets as independent variables, yielded coefficients β_usage ≈ 0.31, β_Mendeley ≈ 0.37, and β_Tweets ≈ 0.09, with an overall R² of 0.36. These results suggest that while all three alternative metrics are positively associated with citation impact, usage and capture are stronger predictors than Twitter activity.

The authors also examined citation ageing patterns. Preprint articles displayed a steep citation growth in the first two years post‑publication, followed by a plateau, whereas non‑preprint articles showed a slower, more linear increase over time. This supports the notion that the “early‑view” advantage of preprints can have lasting effects on cumulative citation counts.

Limitations acknowledged include potential duplication or omission across bibliographic databases, the academic bias inherent in Mendeley readership (which may underrepresent non‑scholarly audiences), and the English‑language dominance of Twitter data, which could skew social‑media metrics. Moreover, the study does not directly assess the quality differential between preprint and final versions, leaving open the question of whether the observed advantages are partly driven by self‑selection of higher‑quality work into the preprint channel.

In conclusion, the empirical evidence presented demonstrates that arXiv preprints confer both “early‑view” and “open‑access” benefits that translate into higher citation counts, greater readership, and more rapid social‑media attention. Usage and capture metrics exhibit a stronger relationship with citations than Twitter mentions, reflecting differing scales and user bases of these indicators. The findings substantiate the role of preprints as a catalyst for faster and broader dissemination of mathematical research, offering a data‑driven rationale for institutions, funders, and publishers to encourage preprint posting as part of the scholarly communication workflow.