Microsimulation Models of the Criminal Justice System Research Paper

This sample Microsimulation Models of the Criminal Justice System Research Paper is published for educational and informational purposes only. If you need help writing your assignment, please use our research paper writing service and buy a paper on any topic at affordable price. Also check our tips on how to write a research paper, see the lists of criminal justice research paper topics, and browse research paper examples.

This research paper provides an introduction to the use of microsimulation models for informing decision making by criminal justice system professionals. Microsimulation models are placed within a framework for understanding computer simulation models. A brief history of the use of simulation models and microsimulation models in the criminal justice system is provided. The processes involved in the development of the Queensland Juvenile Justice Simulation model are described to illustrate the steps required to develop, validate, and use a microsimulation model. Finally, some of the challenges and advantages to using microsimulation models in the criminal justice system are explored.

Overview Of Social Science Microsimulation Modelling

Within the social sciences three broad approaches to computer simulation models have been identified: macrosimulation approach, microsimulation approach, and agent-based modelling approach (Macy and Willer 2002). The development of macrosimulation models (also known as system dynamics models, world models, stock and flow models, dynamic systems simulation modelling) coincided with the first use of computers in university research in the 1960s. These models have their roots in systems of difference and differential equations and are used for understanding the dynamic nature of complex nonlinear systems. The second modelling approach, microsimulation (also known as discrete event simulation modelling), simulates a “macro” system by modelling and simulating the actions and interactions of the systems “micro” units (usually individuals) (Spielauer 2011). The results of the effect of system changes on each unit can then be analyzed at the microlevel or aggregated up to show the overall “macro” effect of change. This type of modelling has its roots in economics, engineering, and operations research. The final approach to simulation modelling, agent-based modelling, was developed during the 1980s coinciding with the advent of the personal computer (Macy and Willer 2002). This type of modelling has its roots in artificial intelligence and computer science. Unlike macroand microsimulation the focus of agent-based modelling is on theoretical development and explanation rather than applied research (Gilbert and Troitzsch 2005).

Social science microsimulation enables the experimentation with a virtual society of thousands of individuals created by a computer. These models can either be static (examine the short-term effect of a policy change) or dynamic (follow individuals over the life course). In static models time has no effect on the individuals modelled. However, dynamic simulation models explicitly include time. For dynamic microsimulation models characteristics associated with the individual units change in response to accumulated experience or the passage of time. Consequently these models enable us to simulate future conditions and project the outcomes of a proposed policy or operational change. By understanding the likely consequences of a proposed change, it is possible to evaluate if the policy change would be beneficial, contributing to the process of policy analysis and implementation. Dynamic microsimulation models allow virtual individuals to be followed over their entire life course adding a whole new dimension to policy analysis.

There are two steps to social science simulation: first, building the computer model and the second is to “play” or “experiment” with the model. To build a microsimulation model, detailed data are required on the individuals in the system. Individuals are then modelled with a number of key attributes (e.g., sex, age, race) and a number of transition probabilities that reflect changes over time. Once the models are built, they are run (multiple times) to establish a baseline scenario. These baseline scenarios reflect expectations of how current conditions will impact future populations, assuming the policies and the system remain constant and the only changes are the underlying demographics.

Simulation scenarios ask the “what if” questions. They are like mini experiments that identify the downstream impact on the system of a proposed change if everything else was held constant. Of course, systems are extremely dynamic and models do not and cannot predict the future. Rather, models provide predictions on the basis of past trends and take into account what is known about a particular system. As such, policy simulation modelling provides decision makers with additional information that would assist them in making rational decisions on the optimal use of scarce resources and improve the accountability of the criminal justice system.

Microsimulation provides a tool for examining the impact of the behavior of individuals on the whole system. The social science literature is rich with statistical micro data. However, most research stops after the estimation of individual processes. As Spielauer (2011) stated, “With a microsimulation model, we go one step further: microsimulation adds synthesis to analysis” (p. 11). Spielauer (2011) identifies three circumstances under which microsimulation is the appropriate simulation approach, when population heterogeneity is important, when behaviors are complex at the macro level but better understood at the microlevel, and when individual histories matter. These circumstances make microsimulation modelling an appropriate modelling technique for policy analysis in the criminal justice system.

History Of Simulation Modelling Of The Criminal Justice System

Microsimulation is widely used in the social sciences for policy analysis. Policy-orientated microsimulation modelling is commonly used in areas such as pensions, tax policy, income and poverty issues, and health policy. Within the criminal justice system microsimulation is not currently widely used; however, this modelling approach was quite prominent in the criminal justice system in the late 1960s (Auerhahn 2008a).

The pioneer of criminal justice system simulation modelling is Alfred Blumstein. Over 40 years ago, he, in his role as Director of President Johnson’s Commission on Law Enforcement and Administration of Justice, introduced techniques from operations research, quantitative modelling, and system perspective and planning to criminal justice (Blumstein 2007). This work formed the basis for the development of the seminal computer-based criminal justice simulation model – the Justice System Interactive Model (JUSSIM) (Belkin et al. 1972). JUSSIM made it possible for the first time to examine how actions in one part of the criminal justice system affected other parts of the system pioneering the use of policy analysis tools in the formulation of criminal justice policy (Nagin 2012). This model represented the criminal justice system as a sequence of stages, starting with crime, leading to arrest and leading through the stages of the criminal justice system, ending with parole and release from parole. A major limitation of the model was the inability to incorporate feedback (recidivism) into the model. This limitation was addressed in the second version of the model JUSSIM2 (Belkin et al. 1973). This model and its variants are still in use in the USA.

Following on from this work simulation models were developed in the UK (Morgan 1985a, b) and in Australia (Crettenden et al. 1983). These early representations of the operation of the criminal justice system did so in great detail allowing for a greater range of policy simulations to be conducted. However, this increase in detail resulted in a more complex model with more assumptions included and more parameters to be estimated. This made these models difficult, time consuming, and expensive to update and maintain, and mostly these models have disappeared (see review Stewart et al. 2004).

One of the most significant findings of the early work on understanding and modelling the criminal justice system was that people in the criminal justice system are likely to end up back in the criminal justice system – the problem of recidivism. This initiated a productive line of research by Blumstein and colleagues examining the criminal careers of offenders. Criminal careers are the characterization of longitudinal patterns of offending and the distribution of these overpopulation subgroups (Blumstein 2002). This research is still very active today (Piquero et al. 2003), and the findings have considerable policy implications for the control of crime including prevention, career modification (deterrence), and incapacitation. Furthermore, this work is the forerunner to developmental and life-course criminology (Farrington 2003), currently the most influential theoretical paradigm in criminology (Cullen 2011).

In recent years there has been a resurgence in criminal justice simulation models for policy analysis. Much of this work is carried out in house and is not published in the academic literature. Examples of work that has been published included Zarkin and colleagues (Zarkin et al. 2005, 2012) who utilized microsimulation to demonstrate the lifetime costs and benefits to society of improving prison-based post release substance abuse programs. Auerhahn (2002, 2007, 2008b) used system dynamic modelling to examine sentencing reforms, the increase of the length of stay, and the impact of aging prison populations on future prison populations. The New South Wales Bureau of Crime Statistics built a stock and flow model of the criminal justice system (Clark and Lind 2003; Lind et al. 2001) and used this model to simulate flows through the New South Wales criminal justice system.

The construction and use of the Queensland Juvenile Justice System Model is described in the next section. This description will provide the reader with a brief introduction to steps involved in building a model. There is very little publicly available documentation on the building of criminal justice microsimulation models. Following this description two published examples of the use of this model for policy analyses are briefly described. The difficulties in maintaining and using the model will then be briefly outlined.

Queensland Juvenile Justice System Model

An example of a criminal justice system microsimulation model is the Queensland Juvenile Justice System (QJJSM). The development of this model is fully explained in a technical report (Stewart et al. 2004). This model was developed in collaboration with the Office of Economic and Statistical Research, Queensland Treasury, to provide a tool for policy makers and legislators to estimate the relative impact of potential changes to the system in the medium term. The model was designed to assess “what if” questions, with the underlying assumption that, apart from the system change being modelled, the only changes to the youth justice system relate to demographic changes. Therefore, the focus of the model was on a comparison over time between the baseline situation (i.e., the current system) and the proposed change to the system.

This model is a parsimonious model that simulates the initiation of new offenders, the commission of specific offences, the decision of the youth court, and reoffending behavior (Fig. 1). The model is a microsimulation model, that is, individual offenders are simulated. This allows for the examination of both the way individuals move though the system and the number of young people at various points in the system (e.g., detention). The outputs of this model are individual-based data that can be summarized and analyzed using standard statistical techniques to examine and explore the medium-term impacts of the intervention programs on the number of young people flowing through the system.

Microsimulation Models of the Criminal Justice System Research Paper

There are a number of leverage points included in the model. These leverage points represent components of the youth justice model where the implementation of a program, policy, or legislative change may result in the reduction of offending. These leverage points were designed to allow the user to add crime prevention, diversion, and post-court intervention programs to the base system. This enabled the comparison of the impact of these interventions with the baseline system.

Keeping the model simple resulted in three benefits. First, the model captured the critical components of both the system (court outcomes and supervision) and individual offender behavior (initiation, desistance, and reoffending) required. Second, the model’s simple structure made easy to explain to policy makers thereby avoiding alienating the model’s user base through over complexity. Third, the structure of the final schema ensued that the data required to parameterize the mode were readily available in the administrative datasets maintained by Queens-land government.

The model was developed in the proprietary microsimulation package Extend. The logic for choosing a proprietary package and this particular package is provided in the technical report (Stewart et al. 2004). Prior to building the model, the administrative data collected by Queensland government to manage the operation of the youth justice system were statistically examined to develop offending rates, offence probabilities, sentencing models, and models of reoffending and desistence. Separate analyses were conducted for each stage of the system and included a series of logistic regressions for the court decisions and survival analyses for reoffending. The details of these analyses are presented in Stewart et al. (2004).

In the model new offenders enter the system from the general population. Based on the analyses of the administrative data, these offenders are probabilistically assigned demographic characteristics (age, sex, Indigenous status, and geographic region). The offence types and court outcomes were reduced to categories that meaningfully reflected statistical differences in behavior and outcomes among the demographic groups. Eight offence types were modelled. These included offences against the person, break enter and burglary, theft and related offences, drug offences, traffic offences, public order offences, property damage, and other offences. Analyses indicated that gender and Indigenous status determined the offending profiles for young offenders and there was no evidence of offence specialization. That is, regardless of the number of previous offences, the probability of a particular offences category remains constant. Consequently, the model assigns probability parameter to each demographic group, corresponding to the offending profiles identified in the data.

To determine the court outcomes (divert from formal order, non-supervised order, community supervised order, detention order suspended (immediate release order), and detention order served), a series of logistic regressions were performed to examine a range of predictive variables (number of prior appearances, Indigenous status, gender, offence type, previous detention, number of previous detention orders, number of offences finalized at appearance, total number of prior finalized offences). These analyses indicated that two variables accounted for the majority of the variation in sentencing (offence type and the number of prior finalized appearances). Probability tables based on these two variables (and gender, due to the simplicity of including it and its significance in some, but not all, of the logistic models) were created and used in the model to assign court outcomes to offenders.

The final, and arguably the most critical parameter, to be modelled was the reappearance of offenders. Young offenders re-appeared in the model if they reoffended before they turned 17 years. The analyses revealed that 65 % of court appearances were by young offender reappearing in the system. Consequently, the reappearance parameters were critical to ensure the model replicated the system. The number of times an individual reappears in the model also determines the number of prior appearances assigned to that individual. This variable was identified as critical in predicting sentencing outcomes. Survival analyses indicated that there were significant differences in the time to reappear between the four demographic groups (gender by Indigenous status). These differences were apparent in the time from first to second appearance, second to third appearance, and third to fourth appearance. After the fourth appearance the survival curves for the four groups were converging (and the numbers in each group were small). Two population Weibull distributions (with a desistance term) were fitted to the 16 categories (gender, indigenous status, and appearance number (1,2,3,4+)) to provide reappearance parameters for the model.

The model was validated by comparing the simulation results for court outcomes for the financial year 2002/2003 with the actual court outcomes for that year. These results were aggregated by court outcome and Indigenous status of the young offender. Only the detention outcomes for both Indigenous and non-Indigenous offenders fell outside the 95 % upper and lower bounds of the simulated values. The simulation model underestimated the number of young people sentenced to detention orders. These numbers are small, show high variability, and therefore are difficult to estimate. Consequently, these underestimations do not invalidate the model. However, the detention numbers produced by the model need to be treated with caution.

Policy Analysis Using The QJJSM

The primary purpose of the QJJSM is the analysis of proposed changes to policy in the youth justice system. The model cannot predict exactly what is going to happen in the future, as there are too many influential factors that cannot be modelled accurately. Instead, the model provides a baseline set of data assuming the currently system behavior will remain stable over the time period modelled, with only the underlying demographics changing (Baseline Scenario). These demographics are obtained from the projected population estimates, which indicate differential population growth between Indigenous and non-Indigenous young people. This baseline model provides a set of standard output that can be used for comparison with the proposed system changes (Policy Scenario).

Once the baseline results have been recorded, the user can include one or more prospective programs at the leverage points. The model is then rerun with the proposed programs included and the relative reduction in the flow of young people through the youth justice system can be examined. Due to the nature of microsimulation models, the resulting data are individual based and consequently can be analyzed using conventional statistical techniques to examine the impact of the proposed intervention both on the system (e.g., the reduction in court appearances or detention orders) or individuals (e.g., the reduction in Indigenous offenders). Furthermore, the model allows some simple cost analysis by incorporating the costs associated with court appearances, supervision of community-based orders, and detention.

Two examples of policy analysis using the Queensland Juvenile Justice Simulation Model have been published (Livingston et al. 2006; Stewart et al. 2008). Each of these will be briefly described.

Modelling An Early Intervention Program

Livingston et al. (2006) presented the results of the introduction of a family-based counselling program aimed at 5to 10-year-olds. This simulated program targets young people who have yet to come into contact with the youth justice system. The results of Farrington’s (1994) metaanalysis were used to provide estimates of the efficacy to parameterize this scenario. This meta-analysis indicated that such programs demonstrate a 12 % reduction in the initiation of offending by young people. Consequently, the model assumed that this program will prevent 12 % of generated new offenders from commencing a criminal career. Furthermore this simulated program was geographically targeted into areas with high populations of young Indigenous Australians.

The results of simulation indicated that immediately after introduction the program there was little impact of the program on the number of young people initiating offending. This was primarily due to the young age of the program participants. However, after a period of 10 years (the life of the simulation), the simulation estimated a reduction of 6.5 % in the number of court appearances by Indigenous young people and a reduction of 1.7 % in court appearances by non-Indigenous young people. This example highlights the benefits resulting from programs that target preadolescents can take a number of years to realise but over the longer term such programs can have a substantial impact on the number of young people appearing in the youth justice system.

Modelling A Youth Justice Conferencing Program

In Australia the use of restorative justice conferencing for youth offenders to divert young people from formal processing in the court has been introduced in all state jurisdictions. The QJJSM model was used to examine the impact of the introduction of youth justice conferencing in Queensland on the overrepresentation of Indigenous young people in the youth justice system (Stewart et al. 2008). Wider utilization of youth justice conferencingwas proposed as a key strategy for reducing the high levels of overrepresentation of Indigenous Australians in the criminal justice system. Young offenders who have pleaded guilty are referred to a youth justice conference by police as a diversion from court or by the courts either in lieu of sentencing or prior to sentencing. Limited research has examined the impact of youth justice conferencing on reappearance in the criminal justice system. Work conducted in New South Wales (Luke and Lind 2002) compared the rate of rearrest for young offenders whose matter was finalized in the youth court with those who attended a conference. They estimated the rate of rearrest for conferences was 15–20 % lower than that for offenders who attended the youth court. This figure was used in the model as a measure of the efficacy of youth justice conferencing.

In 2003 when this simulation was carried out, a pilot of youth justice conference had been completed and the program was about to be rolled out across Queensland. In the policy scenario youth justice conference was equally available to all young people in Queensland regardless of gender or Indigenous status. This scenario was consistent with legislation and anticipated practices. The results of this simulation indicated the introduction of youth justice conferencing had an immediate effect of the number of finalized court appearances for both Indigenous and non-Indigenous young people. By 2011, there was a drop of 12.5 % in court appearances; non-Indigenous young people had a 13.7 % drop and Indigenous had a 10.5 % drop. However, these results need to be interpreted cautiously as over half the estimated reduction results from diversion for the court system, not a reduction in offending. The differential impact of youth justice conferencing is a result of the different offending profiles of Indigenous and non-Indigenous young people. Indigenous young people are younger at their first appearance, offend more frequently, and are less likely to commit property offences that their non-Indigenous counterparts (Stewart et al. 2004). Consequently a diversion program available principally for first time offenders and property offenders will have a greater impact on the number of court appearances and offending behavior of non-Indigenous young people.

The results of this simulation indicate that youth justice conferencing is unlikely to contribute to a reduction in the overrepresentation of Indigenous young people in the criminal justice system. In fact it has the potential to increase the disparity in the ratio of Indigenous to non-Indigenous offenders. These results contribute to the increasing debate about the need to focus outside the criminal justice system to reduce the rates of offending in Indigenous communities. Rather than focusing on criminal justice interventions (such as youth justice conferencing), more progress might be made if the focus was shifted to the underlying causes of aboriginal crime (e.g., substance abuse, family violence, poor school performance, and unemployment).

Current Status Of The QJJSM

Unfortunately, the QJJSM has followed the footsteps of many previous models and is no longer functioning. Not only are microsimulation models extremely resource intensive to build; they also require ongoing resources to maintain and operate them. This model needed to be reparameterized annually with new data to ensure that the model reflected current practices in the system. The model also needed to be updated to reflect changes occurring within the system such as new legislation and new sentencing options. Additionally, computer technology and microsimulation software is constantly changing and this model needed to be updated to take advantage of these new developments. Finally, it was difficult to convince criminal justice policy analysts about the value of simulation modelling. Within Queensland Treasury the model had “champions” who assisted with the initially building of the model. However, across the criminal justice system, policy analysts struggled to understand how they could use the model. The developers identified and tested the different policy scenarios. However, the ultimate aim was to have these policy scenarios generated from within government. Despite best efforts to educate the policy analysts, this did not happen. Consequently government did not continue to resource this model. However, the work in building the QJJSM has resulted in two important outcomes.

First, to build the model it was necessary to gain access to and develop a good understanding of the administrative data held by the various agencies of the Queensland criminal justice system. A critical parameter in building the model was the reappearance of young people in the criminal justice system. To analyze reappearance a longitudinal database was built by linking together administrative data for individuals born in 1983 or 1984, who offended and had contact with the youth justice system (police cautioning and youth court appearances). This work is ongoing and administrative data for these birth cohorts was last linked in 2011 when the individuals were 25/26 years old. The current longitudinal database (1983/84 Queensland Longitudinal Database) includes administrative data on contacts with the child protection system, the youth justice system, the adult court system and administrative data concerning their contacts with the mental health system are currently being negotiated. These data provide the basis of a productive ongoing research program examining the life course of offending, transition and turning points, cost of offending, evaluation of interventions, and geographic distribution of chronic and serious young offenders.

The second major benefit of the modelling program has been the development of strong relationships with influential public servants across the various agencies of the criminal justice system. These relationships have facilitated the research policy nexus, enabling researchers to influence the development of policies and legislation within the Queensland criminal justice system.

The Future Of Microsimulation Modelling Of The Criminal Justice System

In a special issue of Social Science Computer Review, Spielauer (2011) presented a very optimistic future for microsimulation modelling in the social sciences:

Microsimulation is an approach whose time has come. More than half a century after the introduction of microsimulation into the social sciences, the main obstacles to this approach have almost disappeared. Computer power has increased exponentially, the collection individual data has become routine, social scientists are trained in micro data analysis and longitudinal research and the research itself has moved from a macro to a micro approach and is on the way toward a multilevel integration. The life course perspective has become a dominant paradigm and many of the pressing problems we faces are of a nature which makes dynamic microsimulation the most suitable study approach. (p. 10)

Despite this optimistic approach and the acceptance of policy-orientated microsimulation modelling in areas such as pensions, tax policy, income and poverty issues, and health policy, microsimulation modelling is still rare in the criminal justice system (Anderson and Hicks 2011). As Blumstein (2002) concluded, Management of the agencies of the criminal justice system is still far from the model of efficiency one might like, and is still slowly moving into the information technology era (Blumstein 2002, p. 22).

There are two key obstacles to simulation modelling practices becoming an inherent part of criminal justice planning, program, and policy development: the lack of skills among criminal justice policy makers and the politicized nature of the criminal justice system. The first obstacle to the use of simulation models is that criminal justice policy makers and practitioners do not understand simulation technology. These professionals struggle with assumptions required to build a model of a highly complex system, do not understand the various sources of data and analyses required to parameterize the model, and do not have the skills to develop and quantify policy scenarios. These professionals consider simulation models to be hard-to-operate-and-understand black boxes (Spielauer 2011). The traditional training of the criminal justice policy analyst and professional decision makers provides little or no exposure to simulation model-ling. While graduates from criminal justice programs need not have the expertise to develop the models, they should be taught to appreciate the benefits of modelling and to incorporate the outputs of models into their professional practice. Without such training it is difficult for these professionals to develop a professional culture conducive to the use of models and simulations in criminal justice decision making. Without an understanding of the benefits of modelling, it is difficult for governments to justify the financial investments necessary to build, maintain, and operate the models.

The second obstacle to simulation modelling becoming normative practice in the criminal justice system is the highly politicized nature of the system. Arguably, crime and justice issues are the most politicized of public agendas. In recent years, with an increasing public concern about crime and crime rates, we have seen increasingly unsophisticated responses by our politicians to the management and prevention of offending. In these circumstances it is very difficult for researchers to encourage the use of “evidence-based” policy development let alone the use of simulation models. Furthermore, the reactive nature of current policy responses result in fluctuations within the criminal justice system that are extremely difficult to model. Auerhahn (2008a) argues that given the politicized nature of criminal justice policy, simulation modelling enables the consideration of more “radical” policy ideas at very little political or operational cost. Unfortunately, there is limited evidence that these policy ideas are implemented into actual practice in the criminal justice system.

Despite these challenges to the use of microsimulation modelling, three ways in which simulation modelling has the potential to aid decision making within the Australian criminal justice system have been identified. First, and perhaps most importantly, modelling has the potential to lead to a more informed debate about the allocation of scarce public resources. The Australian criminal justice system cost over $13 billion in the 2010–2011 financial year. These costs are increasing dramatically. However, crime victimization data indicates only slight increases in actual crime. In essence, the additional costs to the system can be largely attributed to changes in policing practices, public policy, and legislation, which are often the result of politically driven decisions in response to public concern about crime. This is reflected in criminal justice policies summed up in political catch phrases such as “get tough on crime,” “three strikes and you are in,” “zero-tolerance policing,” and “truth in sentencing” which have resulted in increases in police numbers, prison beds, and more punitive legislation. Indeed, as a result of such policies, the male incarceration rate in Australia has grown from 281 per 100,000 in 2002 to 324 in 2011. However, little attention has been paid to both the initial and cumulative costs of adopting such policies and services over a period of time (Austin et al. 1992).

A second way that policy simulation modelling may aid decision making is that such models can provide information that could assist decisions aimed at providing a more just and equitable criminal justice system. As demonstrated by the use of the QJJSM, microsimulation modeling enables an examination of the differential impact of indigenous status, gender, and geographic locality on young people entering and reappearing in the youth justice system. Currently, Indigenous young people are grossly overrepresented across Australian criminal justice systems. The modelling process facilitates the identification of points in the system where Indigenous young people (and other population groups) are differentially treated. Consequently, the use of modelling has the potential to lead to a more effective, accountable, and equitable criminal justice system.

A third benefit of policy simulation modelling is that it may provide information to decision makers crucial for the effective short-term and long-term planning of the system and the implementation of a whole-of-government approach to criminal justice in Queensland. There is an increasing recognition of the need for a more coordinated criminal justice system. Each section of the criminal justice system has its own policies, data collection strategies, and legislation. However, the complex interdependent nature of the criminal justice system means that any change in one section of the system could have substantial flow on effects to other parts of the system. Statistical models are required in order to determine the consequences of changes in policy and legislation not only for the host agency but also for other agencies in the criminal justice system.

The criminal justice system has much to gain from the use of dynamic microsimulation models. The process of modelling requires a sophisticated understanding and analysis of both the system and the data in the system. The modellers are required to understand what data are collected, why they are collected, what the limitations of these data are, and what gaps there are in the data collection. Simulation modelling transforms these data, which are routinely collected at great expense by the criminal justice agencies, into sophisticated decision support systems. This information has the potential to facilitate policy analysis, planning, and decision making leading to a more equitable and just criminal justice system.

Bibliography:

Allard R (2010) Understanding and preventing indigenous offending. Indigenous justice clearing house brief No 9. Australian Institute of Criminology, Canberra
Anderson RE, Hicks C (2011) Highlights of contemporary microsimulation. Soc Sci Comput Rev 29:3–8
Auerhahn K (2002) Selective incapacitation, three strikes and the problem of aging prison populations: using simulation modelling see the future. Crimin Publ Policy 1:353–388
Auerhahn K (2007) Do you know who your probationers are? Using simulation modelling to estimate the composition of California’s felony probation population 1980–2000. Justice Q 24:28–47
Auerhahn K (2008a) Dynamic systems simulation analysis: a planning tool for the new century. J Crim Justice 36:239–300
Auerhahn K (2008b) Using simulation modelling to evaluation sentencing reform in California: choosing the future. J Exp Criminol 4:241
Austin J, Cuvelier S, McVey A (1992) Projecting the future of corrections: the state of the art. Crime Delinq 38:385–408
Australian Bureau of Statistics (2003) Population projections, Australian, 2002–2101. Australian Government, Canberra
Australian Bureau of Statistics (2011) Prisoners in Australia, catalogue 4517.0. Commonwealth Government of Australia, Canberra
Australian Institute of Criminology (2012) Australian crime: facts and figures: 2011. Australian Institute of Criminology, Canberra
Belkin J, Blumstein A, Glass W, Lettre M (1972) JUSSIM, an interactive computer program and its uses in criminal justice planning. In: Cooper G (ed) Proceedings of the international symposium on criminal justice information and statistics systems. Project SEARCH, Sacramento, pp. 467–477
Belkin J, Blumstein A, Glass W (1973) Recidivism as a feedback process: an analytical model and empirical validation. J Crim Justice 1:7–26
Blumstein A (2002) Crime modeling. Oper Res 50: 16–24
Blumstein A (2007) An OR missionary’s visits to the criminal justice system. Oper Res 55:14–23
Cassidy RG (1985) Modelling a criminal justice system. In: Farrington DP, Tarling R (eds) Prediction in criminology. State University of New York Press, Albany, pp. 193–207
Clark J, Lind B (2003) The New South Wales criminal justice system simulation model: further developments. Crime Justice Bull No 76. New South Wales Bureau of Crime Statistics and Research. http://www. lawlink.nsw.gov.au/bocsar1.nsf/pages/index
Crettenden I, Parker J, Macalpine S (1983) A computer simulation model of the district criminal court of New South Wales. New South Wales Bureau of Crime Statistics and Research, Sydney
Cullen FT (2011) Beyond adolescence-limited criminology: choosing our future – the American society of criminology 2010 Sutherland address. Criminology 49:287–330
Farrington D (1994) Early developmental prevention of juvenile delinquency. Crim Behav Ment Health 4:209–226
Farrington DP (2003) Developmental and life-course criminology: key theoretical and empirical issues – the 2002 Sutherland award address. Criminology 411:221–256
Gilbert N, Troitzsch KG (2005) Simulation for the social scientist, 2nd edn. Open University Press, Berkshire
Gordon MB (2009) A random walk in the literature on criminality: a partial and critical view on some statistical analysis and modelling techniques. Eur J Appl Math 21:283–305
Lind B, Chilvers M, Weatherburn D (2001) Simulating the New South Wales criminal justice system: a stock and flow approach. New South Wales Bureau of Crime Statistics and Research. http://www.lawlink.nsw.gov. au/bocsar1.nsf/files/r50.pdf/$file/r50.pdf
Livingston M, Stewart A, Palk G (2006) A microsimulation model of the juvenile justice system in Queensland. Trends and issues in crime and criminal justice No. 307. Australian Institute of Criminology, Canberra
Luke G, Lind B (2002) Reducing juvenile crime: conferencing versus court. Crime Justice Bull No. 69, New South Wales Bureau of Crime Statistics
Macy MW, Willer R (2002) From factors to actors: computational sociology and agent-based modelling. Annu Rev Sociol 28:143–166
Maltz MD (1996) From Poisson to the present: applying operations research to the problems of crime and justice. J Quant Criminol 12:3–61
McEwen TJ (1992) CJSSIM: Criminal justice system simulation model user manual. Institute for Law and Justice, Alexandria
Mollona E (2008) Computer simulation in the social sciences. J Manag Gov 12:205–211
Morgan PM (1985a) Modelling the criminal justice system. In: Moxon D (ed) Managing criminal justice. HMSO, London, pp 29–45
Morgan PM (1985b) Modelling the criminal justice system. Home office research and planning unit paper 35. Home Office, London
Nagin D (2012) Alfred Blumstein. In: Assad A (ed) The founders of operations research. Springer, New York, pp 707–719
Piquero AR, Farrington DP, Blumstein A (2003) The criminal career paradigm background and recent developments. Crime Justice Rev Res 30: 359–506
Productivity Commission (2012) Report on government services 2011. Australian Government, Canberra
Spielauer M (2011) What is social science microsimulation? Soc Sci Comput Rev 29:9–20
Stewart A, Spencer N, O’Connor I, Palk G, Livingston M, Allard T (2004) Juvenile justice simulation model: a report on the Australian Research Council Strategic Partnerships with Industry Research and Training Grant No C00106983. http://www.griffith.edu.au/data/assets/pdf_file/0009/54846/jmagmodel.pdf
Stewart A, Hayes H, Livingston M, Palk G (2008) Youth justice conferencing and indigenous overrepresentation in the Queensland juvenile justice system: a micro-simulation case study. J Exp Criminol 4:357–380
Weatherburn D, Fitzgerald K, Hau J (2003) Reducing aboriginal over-representation in prison. Aust J Public Adm 62:65–73
Zarkin GA, Dunlap LK, Hicks KA, Mamo D (2005) Benefits and costs of methadone treatment: results from a lifetime simulation model. Health Econ 14: 1133–1150
Zarkin GA, Cowell AJ, Hicks KA, Mills MJ, Belenko S, Dunlap LJ, Houser KA, Keyes V (2012) Benefits and costs of substance abuse treatment programs for state prison inmates: results from a lifetime simulation model. Health Econ 21:633–652