LEAPS-MPS: Development of Models in Spatial Statistics for Complex Policing and Social Science Applications

社会科学中紧迫的研究问题往往需要对空间数据进行分析，而空间数据已变得越来越复杂。本研究旨在通过创新的统计方法更好地理解和模拟这些复杂性，使PI，其他研究人员和社区成员能够更好地回答与公共政策相关的研究问题。PI还将于2023年秋季在卡尔顿学院举办关于警察使用武力分析的统计挑战的论坛。这项工作将有直接的影响，在卡尔顿学院的本科生谁参加这项研究过程中，卡尔顿以外的学生谁参加警务统计论坛，并通过有意与社区组织的伙伴关系，当地明尼阿波利斯社区。拟议的工作将帮助当地社区的人们回答有关其社区治安的具体问题，使他们能够从数据中添加额外的背景到他们的生活经验。该项目将开发新的统计方法，具有三个具体目标：解决和解释空间不确定性的方法，将有界或受约束的参与者纳入空间模型的方法，以及初步空间模型的时空扩展。第一个目标是开发软件来实现一种受约束的空间隐私方法，该方法在保护原始位置的同时保留原始数据集的统计效用。将提出新的方法来分析由于结构化（隐私）和非结构化（地理编码错误或不精确的数据收集）过程的空间不确定性。第二个目标将开发研究方法，以纳入由于结构化（例如，警察管辖区）和推断（区域联盟）边界。第三个目标是将一个新开发的空间模型，这是一个共享的组件模型，用于表征两个点过程之间的关系，一个时空模型，包括时间动力学。这个奖项反映了NSF的法定使命，并已被认为是值得支持的评估使用基金会的智力价值和更广泛的影响审查标准。

Pressing research concerns in the social sciences often require the analysis of spatial data, which has become increasingly complex. This research aims to better understand and model some of these complexities with innovative statistical methodology that will allow the PI, other researchers, and community members to better answer research questions relevant to public policy. The PI also will host a forum at Carleton College in fall 2023 on Statistical Challenges in the Analysis of Police Use of Force. This work will have immediate impacts on the undergraduates at Carleton College who participate in this research process, the students outside Carleton who attend the statistical forum on policing, and the local Minneapolis community through intentional partnership with community organizations. The proposed work will help those in local communities answer specific questions about the policing of their neighborhoods, allowing them to add additional context from data to their lived experiences.This project will develop new statistical methodology with three specific aims: methods to address and account for spatial uncertainty, methods to incorporate bounded or constrained actors into spatial models, and spatio-temporal extensions of preliminary spatial models. The first aim will develop software to implement a constrained spatial privacy method that protects the original locations while preserving the statistical utility of the original dataset. New methodology will be proposed to analyze spatial uncertainty due to both structured (privacy) and unstructured (geocoding error or imprecise data collection) processes. The second aim will develop research methods to incorporate actors that are spatially constrained both due to structured (e.g., police jurisdictions) and inferred (regional unions) boundaries. The third aim will extend a newly developed spatial model, which is a shared component model for characterizing the relationship between two point processes, to a spatio-temporal model, incorporating temporal dynamics as well.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.