Search
Browse By Day
Browse By Time
Browse By Person
Browse By Policy Area
Browse By Session Type
Browse By Keyword
Browse Artificial Intelligence Presentations
Program Calendar
Sign In
Search Tips
One of the biggest drivers of incarceration in Utah is the community supervision population, where close to forty percent of probationers end up in prison, often due to technical revocations. These revocations concentrate in specific neighborhoods where institutional infrastructure is thinnest and access to healthcare, transit, social networks, and environmental standards is most limited. Yet justice agencies lack the tools to identify which barriers drive failure in which neighborhoods, making it difficult to target services and treatment to reach the people and communities who need them most. This paper addresses that gap by adapting small-area estimation methods from Bayesian spatial epidemiology to model and map criminal justice risk at the census-tract level. The analysis develops a spatial policy decision framework that allows criminal justice agencies and cross-system stakeholders to direct resources toward the communities where support is most needed for successful community integration, and where upstream investment in community health, infrastructure, and social connection can prevent justice system contact before it begins. In public health, small-area estimation has transformed policy by replacing broad regional averages with granular neighborhood-level risk profiles that reveal where health risks concentrate and what drives them. This study applies that framework to criminal justice, using geocoded records from the Utah Department of Corrections covering more than 44,000 supervision episodes (2016-2023) and linking them to tract-level indicators from the American Community Survey, the Utah Healthy Places Index, and other health access measures across 689 census tracts. The study uses INLA-BYM2 Bayesian spatial modeling, a small-area estimation method from spatial epidemiology that borrows statistical strength from neighboring geographies to produce stable risk estimates even in low-population tracts. The model identifies which community conditions drive supervision concentration, accounts for spatial dependence among neighboring communities, and generates tract-level risk maps with Bayesian confidence levels that tell policymakers where to invest and how certain the assessment is. The results demonstrate that educational attainment is the strongest protective factor, reducing tract-level risk by 20.6 percent. Median household income reduces risk by 12.5 percent. Social isolation increases risk by 18.7 percent. Diesel particulate matter, an environmental justice indicator, independently elevates risk by 9.6 percent. Half of geographic variation reflects neighborhood spillovers, meaning that the conditions in one community shape outcomes next door. The analysis identifies 128 high-risk tracts across urban, suburban, and rural communities and classifies them into four profiles that require locally tailored strategies: health-dominated, economic opportunity gap, environmental justice, and social infrastructure gap. These findings translate into an operational decision-support platform built in partnership with Utah's Department of Corrections. The platform allows policymakers, health administrators, and community organizations to identify priority tracts, examine which community conditions drive local risk, and match deficits to intervention strategies. Coefficient contribution analysis pinpoints the binding constraint in each neighborhood, enabling service providers and justice agencies to match resources to local barriers. The spatial policy decision framework developed in this paper offers a replicable methodology for any state seeking to direct limited resources toward the communities and conditions where they will matter most.