Doctoral Dissertation Research: Public Beliefs and Responses to Industrial Sites

公众对工业项目的反应决定了健康和环境风险在社会中的分布。它们还影响新技术和新产业的社会接受度、政府监管和经济可行性。本文以水力压裂技术带来的油气生产热潮为背景，研究工业项目风险和收益的替代概念如何构成动员公众反对和支持的结构。首先，这项研究区分了NIMBY(不在我的后院)反对者和意识形态反对者，前者基于感知到的当地影响而动员，后者从更大的一套政治信仰和价值观对该行业做出反应。其次，研究人员考察了对工业选址的支持，这些支持在现有的学术中很少受到关注。该项目引入了公众评论，作为衡量公众意见的新数据来源。该项目对于理解选址和扩大这个快速增长的行业背后的政治动态具有潜在的重要影响。这项研究使用的方法应该广泛适用于其他可以受益于“大数据”的研究领域。美国目前的能源发展带来了新的社会科学，以解释公众对工业选址的反应。这项研究基于居民的意识形态承诺，对公众对工业选址的反应进行了补充解释，以帮助调和关于接近和反对之间关系的不一致的经验发现，这项研究测试了NIMBY(不在我的后院)和补充解释。该项目将审查在伊利诺伊州和纽约州两个州对水力压裂进行监管审查期间提交的9.1万条公众意见。首先，研究人员将绘制一份全面的地图，说明地理上的接近与动员之间的关系，以支持和反对拟议的水力压裂项目。其次，研究人员将把这些评论与社区背景的衡量标准和个人层面的意识形态衡量标准结合起来，在一系列统计分析中评估对工业选址动员的支持和反对的相互矛盾的解释。最后，作为对所提出的理论机制的直接测试，研究人员将使用自然语言处理技术为评论者谈论水力压裂的不同方式编写评论正文。对于工业选址的研究，公众评论数据提供了三个具体的优势：(1)它们提供了对一个行业的反对和支持的行为测量；(2)它们是地理编码的，允许精确测量评论者？S与拟议的工业选址的距离；(3)公众评论的文本提供了前所未有的洞察，让人们了解人们对行业影响的不同概念。分析的结果，特别是邻近关系和政治意识形态的相对影响，可能会对理解其他政策辩论的动态产生影响。最后，拟议的项目将展示以自然语言技术和机器学习概念为基础的创新方法，以利用现有公共记录研究这些动态，并使用严格的多种方法。

Public responses to industrial projects shape the distribution of health and environmental risks within society. They also influence the social acceptance, government regulation, and economic viability of new technologies and industries. This dissertation uses the context of the unfolding boom in gas and oil production enabled by hydraulic fracturing technology to investigate how alternative conceptions of the risks and benefits of industrial projects structure the mobilization of public opposition and support. First, the research distinguishes between NIMBY ("Not in my backyard") opponents who mobilize on the basis of perceived local impacts, and ideological opponents who react to the industry from a larger set of political beliefs and values. Second, the researchers examine support for industrial siting that has received little attention in existing scholarship. The project introduces public comments as a new source of data for measuring public opinion. The project has potentially important implications for understanding political dynamics behind decisions to site and expand a rapidly-growing industry. The methods used for this study should be broadly applicable to other areas of study that can benefit from "big data."Current energy development in the United States brings renewed social science to explain public reactions to industrial siting. This research tests "NIMBY" (not in my backyard) against a complementary explanation of public response to industrial siting based on residents' ideological commitments to help to reconcile inconsistent empirical findings about the relationship between proximity and opposition. The project will examine 91,000 public comments submitted during the regulatory reviews of hydraulic fracturing in two states, Illinois and New York. First, researchers will generate a comprehensive mapping of the relationship between geographic proximity and mobilization for and against proposed hydraulic fracturing projects. Second, the researchers will combine the comments with measures of community context and with individual-level measures of ideology to evaluate competing explanations of mobilization for, and against, industrial siting in a series of statistical analyses. Finally, as a direct test of the proposed theoretical mechanism, the researchers will use natural language processing techniques to code the body of comments for the different ways that commenters talk about hydraulic fracturing. For research on industrial siting, public comments data offer three specific advantages: (1) they offer a behavioral measure of opposition and support of an industry (2) they are geocoded, allowing for precise measurement of a commenter?s proximity to proposed industrial sites, and (3) the text of public comments gives unprecedented insight into the different conceptions that people develop of industry impacts. The results of the analysis, especially the relative effects of proximity and political ideology, may have implications for understanding the dynamics of other policy debates. Finally, the proposed project will demonstrate innovative approaches based on natural language techniques and machine learning concepts to studying these dynamics using existing public records, as well as the use of rigorous multiple methods.