Using Bayesian Mixed Effect Generalized Linear Models to Evaluate Criminological Interventions: An Application to Firearm Seizures during Directed Patrol

PurposeTo provide guidance on the use of Bayesian mixed-effect modeling for inference in criminological applications involving complex longitudinal data, and a demonstration involving police firearm seizures and gun violence.MethodsWe outline a Bayesian strategy for specifying and fitting mixed-effe...

Full description

Saved in:  
Bibliographic Details
Authors: Rydberg, Jason (Author) ; Greene-Colozzi, Emily Ann (Author) ; McGarrell, Edmund F. 1956- (Author) ; Perry, Sean 1968- (Author)
Format: Electronic Article
Language:English
Published: 2025
In: Journal of quantitative criminology
Year: 2025, Volume: 41, Issue: 4, Pages: 599-622
Online Access: Volltext (lizenzpflichtig)
Check availability: HBZ Gateway
Keywords:
Bayesian regression
gun violence
Illegal firearms
Multilevel models
Description
Summary:PurposeTo provide guidance on the use of Bayesian mixed-effect modeling for inference in criminological applications involving complex longitudinal data, and a demonstration involving police firearm seizures and gun violence.MethodsWe outline a Bayesian strategy for specifying and fitting mixed-effects models using the brms package in R, including procedures for prior predictive simulation, model selection, and posterior comparison during sensitivity analyses.ResultsTo demonstrate the utility of the Bayesian approach, we apply it to complex longitudinal data from a police directed patrol intervention in Flint Michigan, estimating the impact of gun seizures on firearm violence using a three-level mixed effects zero-inflated negative binomial regression model. The demonstration suggested a small and fragile reduction in reported gun violence associated with firearm seizures, which was not sufficiently distinguishable from placebo-based sensitivity analyses.ConclusionsBayesian methods are underutilized compared to frequentist approaches in criminological applications. This manuscript outlines an approach to expanding the criminological "toolkit" for mixed-effects modeling that leverages recent advances in software for Bayesian inference.
ISSN:1573-7799
DOI:10.1007/s10940-025-09610-0