Social media platforms like Twitter enable policymakers to communicate their policy preferences directly and provide a bird's-eye view of their diverse policy agendas. In this article, we leverage politicians’ social media data to study political attention using a supervised machine-learning classifier that detects policy areas in individual tweets. We examine how individual diversity and institutional factors affect differential attention to public policy among members of the U.S. Congress. Our novel approach to measuring policy attention builds on work by the Comparative Agendas Project, in order to study members’ political attention in near real-time and to uncover both intragroup and intergroup differences. Using this classifier, we labeled more than one million tweets and found statistically significant differences in both the level and distribution of attention between parties, chambers, and genders. However, these differences were small enough to suggest that other Congressional members’ characteristics are also at play. We explored institutional factors (e.g., committee assignment, caucus), partisan issue preferences (e.g., issue ownership), and the political environment (e.g., partisan issues, confirmations, etc.) that may help explain the patterns of political attention that appear in Congress's tweets.

中文翻译：

---

是什么推动了美国国会议员对 Twitter 的政策关注？

Twitter 等社交媒体平台使政策制定者能够直接传达他们的政策偏好，并提供对其多样化政策议程的鸟瞰图。在本文中，我们利用政治家的社交媒体数据，使用监督式机器学习分类器来研究政治关注度，该分类器可检测单个推文中的政策领域。我们研究了个人多样性和制度因素如何影响美国国会议员对公共政策的不同关注。我们衡量政策关注度的新方法建立在比较议程项目的工作基础上，目的是近乎实时地研究成员的政治关注度并揭示群体内和群体间的差异。使用这个分类器，我们标记了超过 100 万条推文，发现政党、商会和性别之间的注意力水平和分布具有统计学显着差异。然而，这些差异小到足以表明其他国会议员的特征也在起作用。我们探讨了制度因素（例如，委员会分配、核心小组）、党派问题偏好（例如，问题所有权）和政治环境（例如，党派问题、确认等）可能有助于解释出现在国会的推文。