A Review of Selected Applications of Analytics to Criminal Justice Systems

In this section we will review some selected applications of analytics for the design and operation of criminal justice systems that are published in the literature. As compared to other systems, for example in the areas of health care, production, transportation, supply chains, etc., the applications in criminal justice have been relatively sparse. In addition, the applications involving multiple objective methods in criminal justice have been rare.

Much of the research into criminal justice systems involves studying the effects of policies, laws, and procedures, or studying and predicting criminal behavior. This is as opposed to optimization over an entire set of policies, laws, and procedures. Because of space limitations, the review provided here will not be exhaustive.

There are many journals and books which publish research addressing the area of criminal justice. The list of journals include the following: Criminology and Public Policy, Journal of Drug Issues, Journal of Criminal Justice, American Journal of Criminal Justice, Journal of Experimental Criminology, Journal of Criminal Law and Criminology, The Prison Journal, Crime & Delinquency, Journal of Research in Crime & Delinquency, Journal of Quantitative Criminology, Security Journal, Criminal Justice Studies, Criminology, Punishment & Society, Criminal Justice and Behavior, Policing: An International Journal of Police Strategies & Management, Justice Quarterly, Journal of Drug Issues, Federal Probation, and Crime Science.

There are also many University law journals available, such as the Harvard Law Review and the Emory Law Journal.

Science, PLOS ONE, Kybernetes, and Psychiatric Services are also journals which sometimes contain research in criminal justice.

Finally, journals/proceedings which contain publications in operations research and systems, sometimes with papers that address criminal justice include Operations Research', Management Science', INFORMS Journal on Applied Analytics (formerly called Interfaces)', Decision Analysis; Omega; Journal of the Operational Research Society; European Journal of Operational Research; Socio-Economic Planning Sciences; IEEE Transactions on Systems, Man, and Cybernetics; and Proceedings of the Winter Simulation Conference. The INFORMS Journal on Applied Analytics is of note because of its focus on applications as opposed to theory.

Magazines or media organizations of a more general nature which sometimes contain articles on subjects related to criminal justice include The Atlantic, Politico, The Conversation, and Mother Jones. With respect to at least some of these publications, the reader should be aware of a rare bias which can affect the argument made.

Some of the earliest research in the application of OR techniques to the criminal justice system involved the use of simulation to model the entire system, much like the system illustrated in Figure 1.1. An example of a publication representing this research is Blumstein and Larson (1969). Their model represents flows of criminals/defendants from one stage to another of a criminal justice system. One of the features of the model is that it represents the flow of some of the prison-ers/defendants back into the system after they have left it (a feedback loop). The model is illustrated using data collected in the state of California through 1965. The outputs from the model included projections of workloads, costs, and manpower requirements.

The research of Blumstein and Larson and additional early research of a similar type is described in Chaiken et al (1976) and Bohigian (1977). In particular, Bohigian (1977) divides the models reviewed in his paper according to the categories of criminal justice planning models; police patrol models; prosecution models; forensic science models; court operation models; juror management models; corrections, parole, and probation models; juvenile delinquency models; police communication models; and offender tracking models.

Auerhahn (2002) investigated the effects of sentencing reforms (for example, the three-strikes law) in the state of California on the percent of the prison population that is elderly. In a later work, Auerhahn (2004) used simulation to investigate the influence of California’s Substance Abuse and Crime Prevention Act of 2000 on California’s prison population to the year 2020.

Berenji, Chou, and D’Orsogna (2014) used simulation to compare in a general sense rehabilitation programming vs. prison as a policy for nonviolent drug offenders on recidivism.

In an application like that found in many systems which involve the use of processes to accomplish goals, Larson, Cahn, and Shell (1993) used Monte Carlo simulation to improve the arrest-to-arraign-ment process in New York City. Using the simulation, they developed policies that allowed for the average time for arrest-to-arraignment to be reduced from about 40 hours to 24 hours, at an annual savings of $3.5 million.

Most of the applications of simulation mentioned above involve the use of continuous simulation in which the continuous flows of prisoners/defendants from one state (e.g., convicted, paroled, or free on bond) to another are represented. In a different simulation-oriented approach, Merlone, Manassero, and Raffaello (2016) employed agentbased simulation to model the behavior of 200 individual criminals.

A important aspect of simulation is the concept of validation, that is, whether the simulation is an accurate representation of the system being modeled. Berk (2008) provides some guidance in this area for simulation models of criminal justice systems.

Analytical models such as those involving queueing theory and other types of probabilistic representations have also been used to address issues in criminal justice systems. For example, consider the fact that states have different requirements regarding jury size. Nagel (1981) addressed the decision problem of jury size using probabilistic modeling; the issues addressed included the facts that (1) if there are too many jurors, then many guilty defendants will be declared not guilty (called a type I error) and (2) if there are too few jurors, then some innocent defendants will be declared guilty (called a type II error). Gorski (2012) notes that the US Supreme Court has been making intuitive decisions regarding jury size, instead of investigating the situation on the basis of mathematics.

Harris and Thlagarajan (1975) used queueing theory to aid decision makers in the District of Columbia to manage the capacities of their halfway houses. Yablon (1991) also used queueing theory to model prison populations as an aid in capacity planning for prisons.

Caulkins (1993) used some sophisticated mathematical modeling to show that, counter to intuition, a zero-tolerance policy with respect to usage of illegal drugs can actually result in an increase in usage of these drugs. (A similar situation is mentioned in Chapter 4 where it is suggested that adding more police could actually result in more crime.)

Another popular methodology in operations research is optimization. However, as noted by Blumstein (2002), there have been relatively few applications of this methodology to criminal justice systems because of the complex criteria associated with these systems.

This section has discussed only a few of the applications of operations research methodology to criminal justice systems in order to give the reader a flavor for this area. Additional detail can be found in other review papers such as Maltz (1994); Barnett, Caulkins, and Maltz (2000); Blumstein (2007); and Bae and Evans (2019).

 
Source
< Prev   CONTENTS   Source   Next >