Does open access strengthen the link between online attention and citations?

Abstract Altmetric Attention Scores are often proposed as early signals of later citations. We test whether open access (OA) changes this relationship by comparing OA and non-OA articles within the same hybrid journals. Study 1 uses 60,565 articles published mainly in 2018 in 1,275 Springer hybrid journals, with attention recorded in 2021 and citations in 2022–2026. Attention predicted later citations more strongly for hybrid OA articles, made open at publication, than for other articles (Spearman ρ = 0.38 vs 0.19). After exact matching, a gap of 0.06 (95% CI 0.03–0.10) remained; it shrank further once early citations were controlled. Downloads accounted for most of the attention–citation link. A natural experiment, in which national agreements made hybrid OA the default in several European countries, points to causal effects of OA on attention, downloads and citations: in the same journals, articles from these countries received 29% more citations, while their non-hybrid articles were not cited more. The hybrid OA citation advantage declined as journal impact factor rose. In Study 2, covering 7,073 Nature articles, OA brought no citation advantage over closed articles. In Study 3, Altmetric data on Colombian research show that the gap lies in public attention (news, blogs, X), not in Mendeley readership. Hybrid OA was concentrated among authors in high-income countries and top-ranked universities. Altmetric scores are thus a somewhat better proxy for citations when articles are free to read, but a poor substitute for citations in assessment.