Related Experiment Video
Updated: Jan 9, 2026

Simulation of Human-induced Vibrations Based on the Characterized In-field Pedestrian Behavior
Published on: April 13, 2016
Kernel Density Bandwidth Specification in Neighborhood Violence Prevention Research.
Hiwot Y Zewdie1, Nicole Asa2,3, Ali Rowhani-Rahbar2,3
1Department of Epidemiology, School of Public Health, University of Washington, Seattle, WA, USA. hzewdie@uw.edu.
Data-driven bandwidth selection in kernel density estimation (KDE) for place-based violence prevention can distort results. Researchers must carefully choose KDE parameters to accurately detect intervention effects on crime.
Area of Science:
- Criminology
- Geospatial Analysis
- Public Health
Background:
- Place-based interventions are used to reduce violence.
- Kernel density estimation (KDE) is a method for analyzing geospatial point data, such as crime incidents.
- The effectiveness of KDE is influenced by the bandwidth parameter, which determines the smoothing extent of the data.
Purpose of the Study:
- To quantify the variability of bandwidths selected using data-driven methods in KDE.
- To examine how different bandwidth sizes impact the estimation of simulated place-based intervention effects on violence.
Main Methods:
- Utilized violent crime data from Philadelphia (2013-2023).
- Calculated bandwidth sizes using two default data-driven criteria for crime-year combinations.
- Simulated a hypothetical intervention with a known effect and analyzed its detectability across various bandwidths using Poisson regression.
Main Results:
- Bandwidth sizes varied significantly based on the data-driven method, crime type, and year, ranging from 45.9 to 48,450 feet.
- Simulated intervention effects were only detectable within bandwidths of 200 to 2900 feet.
- Larger bandwidths led to inaccurate conclusions, suggesting the intervention was ineffective or even crime-increasing.
Conclusions:
- Data-driven bandwidth selection in KDE can obscure or distort the true effects of place-based interventions.
- Researchers must critically and transparently select KDE parameters for accurate violence prevention research.
- Appropriate bandwidth selection is crucial for reliably assessing the impact of place-based violence reduction strategies.
Related Concept Videos
Distribution and Dispersion
Relation Between the Distributed Load and Shear
Relationship Formation
Density
Elastic Collisions: Case Study
Cluster Sampling Method
To choose a cluster sample, divide the population into clusters (groups) and then randomly select some of the clusters. All the members from these clusters are in the cluster sample. For example, if you randomly sample four departments from your...

