Doctoral Dissertation Research: Residential Segregation Measures and Their Spatial Properties

居住隔离是一种固有的空间现象，因为它衡量的是一个地区内不同类型的人的隔离。隔离的空间性质导致了三个截然不同但相互关联的挑战，这应该由经验科学家来解释。首先，大多数研究人员使用公共来源提供的数据，随之而来的是对行政边界不一定优化以进行种族隔离研究的担忧。其次，“社区”通常不是有围墙的飞地。这导致了商店、学校和其他共享地点跨社区边界的住宅互动。第三，居住隔离捕捉到个人的社会-空间相互作用，这意味着必须定义一些空间相互作用函数，以便能够衡量隔离。传统的隔离措施假设行政区域等同于社区，忽略了跨社区相互作用的潜力，因此隐含地定义了以行政边界为终点的空间相互作用函数。在隔离测量的发展过程中，已经取得了一些直接应对这些挑战的进展。面对这些挑战，本研究将进一步推进分离测量方法的发展。因此，它首先使用蒙特卡罗方法来研究潜在的空间特性如何在测量的隔离的大小中表现出来。然后，它使用359个大都会统计区的数据来更深入地研究行政边界对测量的种族隔离的影响。提出了一种建立随机人口普查区域的创新方法，以检验实际人口普查区域是否偏离随机模式。这种方法还提供了一个新的隔离测量框架，减少了行政边界对测量的隔离的影响。最后，提出了一类新的种族隔离措施，它结合了种族隔离的两个方面：社区内的人与社区之间的关系。通过在种族隔离文献中第一次平等地对待这些不同的概念，这个项目将对总体上的种族隔离衡量标准，特别是美国的城市体系提供新的了解。虽然种族隔离的法律基础设施几十年前就被拆除了，但它的痕迹仍然存在。由于种族隔离仍然是社会的一部分，重要的是研究人员和政策制定者拥有最好的工具来衡量它。这项研究对衡量挑战和现有方法进行了批判性的关注，并提出了创新的新方法，以解决这一关于社区健康的关键指标缺失和代表性不足的方面。对方法的关注将有助于隔离理论和实证研究。最后，这项工作将利用开放源码软件社区，并为其做出贡献。作为这项工作的结果而开发的新方法和措施将通过开放源码项目PySAL(Python空间分析库)交付给更广泛的研究社区。

Residential segregation is an inherently spatial phenomenon as it measures the separation of different types of people within a region. The spatial nature of segregation results in three distinct yet interrelated challenges that should be accounted for by the empirical scientist. First, most researchers utilize data provided by public sources, and with this comes the concern that administrative boundaries are not necessarily optimized for segregation research. Second, "neighborhoods" are generally not walled enclaves. This results in residential interaction across neighborhood boundaries in shops, schools and other shared locations. Third, residential segregation captures the socio-spatial interaction of individuals meaning that some spatial interaction function must be defined to allow for segregation to be measured. The classic segregation measures assumed administrative areas to be equivalent to neighborhoods, ignored the potential for interaction across neighborhoods and therefore implicitly defined a spatial interaction function that ended at the administrative boundary. The evolution of segregation measurement has seen a number of advances that directly address these challenges. This research will further advance the methods of segregation measurement in the face of these challenges. As such it first uses a Monte Carlo approach to look at how underling spatial properties could manifest themselves in the magnitude of measured segregation. It then uses data on 359 Metropolitan Statistical Areas to delve deeper into the impact of administrative boundaries on measured segregation. An innovative approach is proposed to build random census tracts to test whether actual census tracts deviate from a random pattern. This approach also offers a new framework of segregation measurement that lessens the effect of administrative boundaries on measured segregation. Finally, a new class of segregation measures is proposed that marries two aspects of segregation: relationships of people within neighborhoods and those between neighborhoods. By treating these distinct concepts as equals for the first time in the segregation literature, this project will shed new light on segregation measurement in general, and on the U.S. urban system in particular.While the legal infrastructure of segregation was pulled down decades ago, its vestiges remain. Since segregation remains a part of society, it is important that researchers and policy makers have the best tools available for its measurement. This research takes a critical eye to measurement challenges and existing approaches, and proposes innovative new methods that address missing and underrepresented aspects of this critical metric on the health of communities. The focus on methods will contribute to both segregation theory as well as empirical research. Finally, this work will both utilize and contribute to the open source software community. New approaches and measures developed as a result of this work will be delivered to the wider research community through the open source project PySAL (Python Spatial Analysis Library).