Self-promotion of research papers on social media is ubiquitous but not exercised to the same extent by every scholar. It is unclear whether there are gender differences in the frequency of self-promotion or the benefit it yields for individuals. Here, we examine differences in women's and men's scholarly self-promotion using 23 million Tweet mentions of 2.8 million research papers published between 2013 and 2018 by 3.5 million authors. Our analysis shows that women are significantly less likely (27\%) than men to promote their papers, even after controlling for a number of important factors, including publication year, journal impact, affiliation rank, author productivity, number of citations, authorship position, number of coauthors, and research topics. In addition, women's underrepresentation on Twitter only explains a small portion of the observed gender difference in self-promotion, as the disparity exists even among authors active on Twitter. The magnitude of the gender gap is more strongly associated with papers' journal impact factor than with authors' affiliation rank, previous productivity, or academic discipline. In particular, men are 78\% more likely than comparable women scholars to self-promote papers published in journals with very high impact factor (IF $\geq$ 40), whereas the difference is only 33\% for papers in low-impact score journals (IF $\leq$ 5). Furthermore, we find that women face a ``dilemma'' in online science dissemination -- while they promote their research less often than men on social media, they risk receiving less of a boost in attention than men if they do self-promote. Our findings offer the first large-scale evidence for a gender gap in scholarly self-promotion online and show the circumstances under which the gap is most substantial, helping inform policy to mitigate discrepancies in visibility and recognition.

翻译：研究论文在社交媒体上的自我宣传虽已十分普遍，但不同学者对此的践行程度并不相同。目前尚不清楚自我宣传频率及其带来的收益是否存在性别差异。本研究通过分析2013至2018年间350万作者发表的280万篇论文在推特上的2300万次提及数据，考察了男性和女性学者在学术自我宣传方面的差异。分析表明，即使在控制出版年份、期刊影响因子、机构排名、作者生产力、被引次数、作者排序、合著者数量以及研究主题等多项重要因素后，女性学者宣传自己论文的可能性仍显著低于男性（低27%）。此外，女性在推特上的代表不足仅能解释观察到的性别差异中的一小部分——即使是在推特活跃用户群体中，这种差异依然存在。性别差异的幅度与论文的期刊影响因子关联性最高，而非作者机构排名、过往生产力或学术领域。具体而言，相较于同类女性学者，男性学者比女性更倾向于宣传发表在极高影响因子（IF≥40）期刊上的论文，其可能性高出78%；而对于低影响因子期刊（IF≤5）的论文，该差异仅为33%。研究还发现女性在科学成果在线传播中面临"困境"——她们不仅比男性更少在社交媒体上宣传自己的研究成果，即便主动推广，所获得的关注度提升也低于男性。本研究首次通过大规模证据揭示了学术自我宣传在线行为的性别差异，阐明了差异最显著的条件，为制定政策以缓解学者可见度与认可度不均衡提供了重要依据。