Testing the Validity of a Queueing Model of Police Patrol

This paper describes efforts to validate a multiple car dispatch queueing (MCD) model of police patrol operations using New York City data. The MCD model was designed for use in a computer system that has been disseminated to many police departments in the U.S. to help planners allocate patrol cars among precincts. It has also been used to evaluate specific changes in patrol policy in New York. We define validation as a series of hierarchical procedures ranging from tests of mathematical correctness to evaluations of model robustness. We discuss the difficulties and limitations of assessing the validity of a model of a loosely managed system in which human behavior is central and in which controlled experiments cannot be performed. Focusing on specific uses, we conclude that the MCD model is a good although imperfect description of patrol operations in New York and is a potentially useful planning tool for many other urban police departments.