The Problem
In some communities in the United States, just getting to and from school poses a complex optimization challenge that thousands of children need to solve every day. Children who need to walk to school want to take the short route, but they instinctively avoid danger zones: parts of their neighborhood where shootings occurred recently, or where they are likely to encounter people who might become involved in shootings. This is a problem our children shouldn’t need to contend with, but some of them – disproportionately minority children – have no choice in the matter. This problem has numerous downstream effects related to school absence as well as chronic stress and anxiety, not to mention physical injury. It is a classic example of structural socioeconomic challenges leading to health disparities in minority populations that absolutely cannot in any sense be blamed upon the affected children. Short of eliminating firearms and people willing to use them, what can we do to help children cope with this grossly unfair burden?
Our Approach
Working with Dr. Jonathan Jay of Boston University, we are building a spatial-temporal agent-based model of neighborhoods around middle schools in parts of Detroit. We are using interview and observational datasets to understand how children reason about getting to and from school each day, how they influence one another in their decision processes, and how they react to interventions that school systems might try. These interventions include:
- More bussing
- Adult guards posted at strategic locations
- Adult escorts through danger zones
- Information to guide children’s decisions
Some of these interventions are more costly than others and cost matters in underserved communities. Yet experimenting with children’s safety seems outrageously unethical. Therefore, we use the computational simulation to ethically experiment, creating cost-benefit analyses to guide the hard decisions school systems face when trying to keep children safe and healthy.
Methodologies Used
- Agent-Based Modeling (ABM) to express the behavior of children, including decision processes and communication patterns
- Spatial-temporal modeling to express the routes children choose and the mutating patterns of danger zones
- System Dynamics to model cumulative public health impacts over time
- Parameter Tuning & Scenario Testing based on real-world datasets from experts studying this problem
- Sensitivity Analysis to evaluate robustness of solutions across different community types and school systems
Findings & Impact
This project is in its early stages so there are no findings, yet, but we expect the impact to be valuable decision support for school systems trying to manage this problem on behalf of the children and families for whom they have responsibility.
Status
- Phase: Model design
- Next Step: Model build
- Lead Investigator: Dr. Jonathan Jay
- Collaborators: NexusSIM team, Boston University, Detroit school systems