Collaborative Research: Mass Media and Representative Democracy

这个项目考察了大众媒体在代议制民主政府运作中的作用。它关注媒体内容中政策线索的频率、准确性和清晰度。它还探索了公众如何利用这些线索来了解他们对美国政策的偏好。调查人员从三个重要命题开始。首先，一个反应迅速的民主公众只需要对政策和政策变化有基本的了解。其次，必要的信息(在某些领域)很容易在媒体内容中获得。第三，公民会注意到这些线索，并相应地调整他们的政策偏好。这个项目测试这些主张，利用不同政策领域的媒体报道的差异，利用该国不同地区的多个新闻媒体，了解意见与政策联系的差异。调查人员假设，在一些领域，媒体报道提供了对政策的相当准确的看法，而另一些领域则缺乏媒体内容。公众对政策的认知和偏好的准确性应该相应地受益或受损。该项目对代议制民主如何(以及何时和为什么)运作提供了一种独特的、由“大数据”驱动的调查。技术摘要该项目的核心是一个自动化的内容分析数据集，包含过去35年来涉及一系列支出领域的数百万条关于美国公共政策的新闻故事和电视新闻文字稿。新闻内容来自全文数据库(主要是Lexis-Nexis)，并使用词典和机器学习方法进行分析。通过人工编码验证了该方法的可靠性和有效性。这些内容分析数据与现有的预算时间序列和民意调查一起进行了审查。分析首次大规模探索了哪些政策信息会出现在媒体报道中，哪些不会出现。这将允许团队评估个人的政策偏好可能如何根据他们消费的与政策相关的媒体内容而变化。作为评估这一影响的一部分，该小组将对媒体对八个政策领域的政策报道和实际政策进行直接比较。这些结果为媒体在代议制民主中发挥的作用提供了一个新的独特视角，特别是在公众对政策的反应、跨问题领域和随时间推移方面。

General AbstractThis project examines the role of mass media in the functioning of representative democratic government. It focuses on the frequency, accuracy, and clarity of policy cues in media content. It also explores how the public uses these cues to inform their preferences for policy in the US. The investigators begin with three important propositions. First, a responsive democratic public requires only basic levels of knowledge about policy and policy change. Second,the necessary information is (in some domains) readily available in media content. Third, citizens pick up on these cues and adjust their policy preferences accordingly. This project tests these propositions, leveraging differences in media coverage across policy areas drawing upon multiple news outlets in different regions of the country to understand variation in the opinion-policy link. The investigators hypothesize that there are areas where media coverage provides a reasonably accurate view of policy, and others in which media content is lacking. The accuracy of public perceptions of, and preferences for, policy should benefit or suffer accordingly. This project provides a unique, "big" data-driven investigation into how (and when, and why) representative democracy works. Technical AbstractThe centerpiece of this project is an automated content-analytic dataset comprised millions of news stories and television news transcripts on US public policy, across a range of spending areas, over the past 35 years. News content is drawn from full-text databases (primarily Lexis-Nexis) and analyzed using both dictionary and machine-learning approaches. The reliability and validity of the method is tested through human coding. These content-analytic data are examined alongside existing budgetary time series and opinion polling. Analyses offer the first large-scale exploration into the kinds of policy information that do, and do not, appear in media coverage. This will allow the team to assess how the policy preferences of individuals might change depending on the policy-relevant media content they consume. As part of the assessment of this influence, the team will provide a direct comparison of media coverage of policy and actual policy across eight policy areas. Results provide a new and unique view into the role that media play in representative democracy, particularly in public responsiveness to policy, across issue areas and over time.