FUSE: Fast Unbiased Small-area Estimation for Timely Measurement of Community Perceptions of Policing and Public Safety

Community perceptions of policing and public safety are highly localized and rapidly evolving. Given the increased scrutiny of law enforcement practices in many areas and the need to assess and refine policies and training programs to address social justice issues, there is an urgent need for cost-effective strategies which are capable of measuring such perceptions in a timely manner and with a high degree of spatial detail. Accurate and timely information from citizens about the performance of their local law enforcement is crucial in order for departments to progress and evolve with their everchanging community composition and needs.
In this project, we propose a three stage Fast Unbiased Small-area Estimation (FUSE) approach, which leverages multiple data sources and statistical techniques to meet this goal. In stage one, we employ stratified sampling and cost-effective panel surveys to quickly collect perception data from a diverse population in the target area. In the second stage, we integrate panel survey data with census Public Use Microdata Sample (PUMS) data and use mass imputation to provide unbiased perception estimates at Public Use Microdata Area (PUMA) level. In the third stage the perception estimates are downscaled to the census block group level using the small area estimation technique SPREE. The methodology will be implemented and applied to panel data collected from five cities—Los Angeles, Washington DC/Baltimore, Minneapolis, Memphis, and Chicago. These cities offer geographic breadth and diverse racial composition. Additionally, we will examine multiple panels for panel effects, and study the local and national spatial variations in community perception and their correlation with factors such as race and socioeconomic status. Address-based probability surveys will be used to collect statistically representative data for validation of the FUSE methodology. Finally, the FUSE methodology will be packaged into statistical software to facilitate easy implementation and wide dissemination, with the overarching goal of building better relationships between the police and the citizens they serve. CA/NCF