Transaction Search Form: please type in any of the fields below.
Date: November 22, 2024 Fri
Time: 11:49 am
Time: 11:49 am
Results for prediction
18 results foundAuthor: Austin, James Title: Reliability and Validity Study of the LSI-R Risk Assessment Instrument Summary: The Pennsylvania Board of Probation and Parole (PBPP) selected the Level of Service Inventory-Revised (LSI-R) instrument as its risk classification tool because it introduces dynamic and more current factors into the risk assessment process, beyond the conventional use of static criminal history and demographic factors. The LSI was developed in the late 1970s in Canada through a collaboration of probation officers, correctional managers, practitioners and researchers. The LSI-R is comprised of 54 static and dynamic items across ten sub-scales. While the LSI-R has been researched extensively in other jurisdictions, its reliability and validity specifically for Pennsylvania’s offender population had not yet been tested. Of particular interest is Pennsylvania’s decision to use the LSI-R as a component of its parole decision making guidelines; heretofore, the LSI-R has been used to identify the appropriate level of supervision for probationers and parolees already residing in the community. In this study, the LSI-R’s relevance and usefulness as a decision making tool as applied to an incarcerated population is a key line of inquiry. The Pennsylvania Commission on Crime and Delinquency (PCCD) contracted The Institute on Crime, Justice and Corrections (ICJC) at The George Washington University to conduct a reliability and validation study using the LSI-R scores and recidivism data. The following report summarizes the ICJC’s findings. This project consists of two segments: an assessment of the inter-rater reliability in scoring the LSI-R, and the validation of the LSI-R’s statistical association with recidivism. The reliability assessment was conducted by selecting a sample of 120 prisoners who were scored on the LSI-R on two separate occasions by two independent PBPP institutional staff. The results of the initial reliability test showed that most of the LSI-R scoring items did not meet a sufficient level of reliability. Consequently, a second reliability test was made in September 2002 on another sample of 156 prisoners to determine if the reliability rates could be improved. The validation assessment entailed examining recidivists (for the purposes of this study, arrests, detentions, absconders, and returns to prison are considered recidivists) of approximately 1,000 prisoners who were released from nine LSI-R test facilities in 2001. For each of these prisoners an LSI-R form was completed. The follow-up period was for 12 months, which allowed the researchers to determine which items were associated with recidivism within that period. Details: Washington, DC: Institute on crime, Justice and Corrections at The George Washington University, 2003. 23p. Source: Internet Resource: Accessed November 9, 2010 at: www.portal.state.pa.us Year: 2003 Country: United States URL: Shelf Number: 120273 Keywords: ParolePredictionProbationRecidivismRisk AssessmentRisk Management |
Author: Jain, Sonia Title: The Power of Developmental Assets in Building Behavioral Adjustment Among Youth Exposed to Community Violence: A Multidisciplinary Longitudinal Study of Resilience Summary: Researchers and practitioners have repeatedly noted substantial variation in the behavioral functioning of youth exposed to community violence. Several studies across fields have documented the detrimental effects of exposure to violence, while other studies have considered how developmental assets promote positive youth development. However, few have examined the lives of the many youth who demonstrate resilience (that is, positive adjustment despite risk) and hardly any have examined how developmental assets may shape resilient trajectories into adulthood for youth exposed to violence. What resources and relationships can high-risk youth leverage to tip the balance from vulnerability in favor of resilience? We used generalized estimating equations, a multivariable technique appropriate for longitudinal and clustered data, to examine multilevel longitudinal data from 1,114 youth ages 11-16 from the Project on Human Development in Chicago Neighborhoods (PHDCN). We considered whether baseline family, peer and neighborhood-level protective factors predicted behavioral adjustment 3-7 years later, among youth who were victims of, witnesses of, or unexposed to violence, controlling for individual and neighborhood-level risks. Behavioral adjustment varied across waves and by exposure to violence. In the short-term, being a victim was associated with increased aggression and delinquency. In the long-term, though, both victims and witnesses to violence had higher odds of behavioral adjustment. Family support, friend support and neighborhood support, family boundaries and collective efficacy had protective effects, and family support, positive peers, and meaningful opportunities modified the effect of exposure to violence to increase the odds of behavioral adjustment over time. Policies, systems and programs across sectors that help nurture these specific supports and opportunities can promote positive behavioral trajectories and resilience into adulthood among urban youth exposed to community violence. Details: San Francisco: WestEd Health and Human Development Program, 2012. 72p. Source: Internet Resource: Accessed April 2, 2012 at: https://www.ncjrs.gov/pdffiles1/nij/grants/237915.pdf Year: 2012 Country: United States URL: https://www.ncjrs.gov/pdffiles1/nij/grants/237915.pdf Shelf Number: 124795 Keywords: Behavior ModificationCommunities and CrimePredictionViolenceViolent CrimeViolent Juvenile Offenders |
Author: Fleck, Angela Marie Title: Is Institutional Sexual Misconduct Predictive of Sexual Recidivism Amongst Male Sex Offenders? Summary: There has been a large body of research conducted on establishing a valid set of predictors of sexual offender recidivism in the past 20 years. However, despite findings that indicate that prior history of sexual offenses serves as a primary predictor of sexual offense recidivism, there has been little focus on the impact of institutional sexual misconduct on sexual offense recidivism rates. This study aimed to investigate the relationship between institutional sexual behavior and sexual offense recidivism rates amongst a sample of male offenders who received a sexual misconduct report while incarcerated and/or was convicted of a sexual offense. Additionally, this study explored whether instances of institutional sexual misconduct added to the variance accounted for by actuarial measures commonly used in Sexually Violent Predator Civil Commitment evaluation procedures. Results revealed that there is little association between sexual offense recidivism rates and receipt of institutional sexual conduct reports unless an offender is issued multiple sexual conduct reports during the same period of incarceration. Additionally, the actuarial measures used in the study were not found to be predictive of sexual offense recidivism. Implications for conducting Sexually Violent Predator Civil Commitment evaluations, identifying institutional sexual offender treatment needs, and identifying community supervision practices are discussed, and future research directions are proposed. Details: Milwaukee, WI: Marquette University, 2011. 123p. Source: Internet Resource: Dissertation: Accessed April 3, 2012 at: http://epublications.marquette.edu/cgi/viewcontent.cgi?article=1128&context=dissertations_mu Year: 2011 Country: United States URL: http://epublications.marquette.edu/cgi/viewcontent.cgi?article=1128&context=dissertations_mu Shelf Number: 124806 Keywords: PredictionPrisoner MisconductRecidivismSex Offenders |
Author: Baradaran, Shima Title: Race, Prediction & Discretion Summary: Many scholars and political leaders denounce racism as the cause of disproportionate incarceration of black Americans. All players in this system have been blamed including the legislators who enact laws that disproportionately harm blacks, police who unevenly arrest blacks, prosecutors who overcharge blacks, and judges that fail to release and oversentence black Americans. Some scholars have blamed the police and judges who make arrest and release decisions based on predictions of whether defendants will commit future crimes. They claim that prediction leads to minorities being treated unfairly. Others complain that racism results from misused discretion. This article explores where racial bias enters the criminal justice system through an empirical analysis that considers the impact of discretion and prediction. With a close look at the numbers and consideration of factors ignored by others, this article confirms some conventional wisdom but also makes several surprising findings. This article confirms what many commentators have suspected—that police arrest black defendants more often for drug crimes than white defendants. It also finds, contrary to popular belief, that there is little evidence to support the belief that drugs are linked to violent crime. Also, judges actually detain white defendants more than similarly-situated black defendants for all types of crimes. The important and surprising findings in this article challenge long-held conventions of race and help mitigate racial disparity in criminal justice. Details: Unpublished paper, 2012. 64p. Source: Internet Resource: Accessed April 9, 2012 at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2035064 Year: 2012 Country: United States URL: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2035064 Shelf Number: 124888 Keywords: DiscretionPredictionRacial BiasRacial Disparities |
Author: U.S. Department of Defense. Defense Science Board Title: Task Force Report: Predicting Violent Behavior Summary: This report conveys the findings and recommendations of the Defense Science Board (DSB) Task Force (TF) on Predicting Violent Behavior. This study was chartered and co-sponsored by the Under Secretary of Defense for Acquisition, Technology, and Logistics (USD(AT&L)) and the Under Secretary of Defense for Policy (USD(P)). This DSB study is one of several reviews that resulted from the killings that took place on November 5, 2009 at the Fort Hood, Texas Soldier Readiness Center, and is submitted in response to the Terms of Reference (TOR) of May 21, 2011. The overall conclusions of the Task Force are the following: Mass-casualty attacks are high consequence but very low-incidence. o However, threats of targeted violence are relatively numerous. There is no silver bullet to stop ALL targeted violence. o There is no effective formula for predicting violent behavior with any degree of accuracy. PREVENTION should be the goal rather than PREDICTION. o Good options exist in the near-term for mitigating targeted violence by intervening in the progression from violent ideation to violent behavior and by creating contexts that minimize alienation or isolation. In the near-term, professional threat management as practiced by law enforcement-led Threat Management Units (TMUs) offer effective means to help prevent targeted violence. o TMUs have been widely deployed, with operational success in the private sector, academia, and elsewhere in government – but not across the Department of Defense (with the exception of the Navy Criminal Investigative Service (NCIS)). o The Department of Defense (DoD) must implement threat management standards of practice, with an emphasis on low footprint, high impact TMUs that largely utilize existing resources. Improved information sharing – considering appropriate accommodation for privacy and free religious practice – is a vital enabler of effective threat management. Science and Technology (S&T) shows some promise as an aid to threat management. o Near-term S&T efforts should focus on conducting rigorous case studies and instituting resilience training. o These case studies should include clinical medical, psychological and behavioral indicators as research better defines their relevance and precision. o Over the long-term, screening technology related to biomarkers has potential. Details: Washington, DC: Department of Defense, 2012. 104p. Source: Internet Resource: Accessed November 27, 2012 at: http://www.acq.osd.mil/dsb/reports/PredictingViolentBehavior.pdf Year: 2012 Country: United States URL: http://www.acq.osd.mil/dsb/reports/PredictingViolentBehavior.pdf Shelf Number: 127008 Keywords: PredictionThreat ManagementViolence (U.S.)Violent CrimeViolent Offenders |
Author: McGregor, Catherine Title: Youth Offenders Risk Identification (YORI): A Screening Tool for Youth Offenders in Western Australia Summary: The present study arose from early discussions between representatives of the Department of Corrective Services (DCS), Youth Justice and the School of Law and Justice at Edith Cowan University (ECU) in August 2009. Investigators from the School of Law and Justice were already involved in a study with DotAG within the children court aimed at identifying the correlates and risk factors involved in youth offending in Western Australia using a content analysis of existing court reports at five year intervals from 1994 to 2009. This project was known as Uncouth Youth? Building a profile of juvenile offenders in Western Australia ("Uncouth Youth"). Discussions with DCS identified an urgent need to develop a brief, valid and user-friendly tool to assist in targeting services towards youth offenders at the greatest risk of re-offending. A valid and reliable screening tool would allow for the triaging of young offenders coming into contact with DCS. This triage system would operate in the same way as those in hospital emergency departments by identifying those young people who should undergo more intensive intervention. There are a number of available instruments designed to identify young offenders at risk of re-offending and to help guide the selection of appropriate interventions aimed at reducing that risk. However, these instruments are generally too lengthy and detailed for standard operational use by busy frontline staff coming into contact with substantial numbers of young offenders and there is a clear need for a more user-friendly tool. Details: Perth: Edith Cowan University, 2010. 29p. Source: Internet Resource: Accessed April 21, 2014 at: https://www.ecu.edu.au/__data/assets/pdf_file/0006/83661/Young-Offender-Risk-Identification-YORI-Report.pdf Year: 2010 Country: Australia URL: https://www.ecu.edu.au/__data/assets/pdf_file/0006/83661/Young-Offender-Risk-Identification-YORI-Report.pdf Shelf Number: 132093 Keywords: Juvenile Delinquency PreventionJuvenile Offender RecidivismJuvenile OffendersPredictionReoffending |
Author: Baird, Chris Title: A Comparison of Risk Assessment Instruments in Juvenile Justice Summary: Juvenile justice service staff began exploring the use of actuarial risk assessments that classify offenders by the likelihood of future delinquency with earnest in the 1970s, but actuarial risk assessments have been used by public social service agencies in the United States since 1928. The value and utility of a valid, reliable, and equitable risk assessment within a broader practice reform effort was made clear to justice agencies in 1998 when the Office of Juvenile Justice and Delinquency Prevention (OJJDP) published the Comprehensive Strategy for Serious, Violent, and Chronic Juvenile Offenders. OJJDP's reform effort illustrated how juvenile justice agencies could better ensure the effectiveness and appropriate targeting of services by implementing both an actuarial risk assessment to accurately, reliably, and equitably classify youth by the likelihood of future delinquency and an equally effective needs assessment to identify an intervention and treatment plan tailored to an individual's needs. This approach built upon the efforts of the National Institute of Corrections' Model Probation/Parole Management Project that combined actuarial risk assessment, individual needs assessment for effective treatment planning, regular reassessments of risk and needs and risk-based supervision standards, and workload-based budgeting. Other models of risk assessment were introduced over subsequent decades, and researchers began categorizing and comparing them as generations of risk assessments. The first generation of risk assessments were not actuarial- individual workers assigned risk levels without the aid of actuarial instruments. Generation 2 instruments were statistically derived, but relied heavily on static criminal history factors to assess risk. They tended to be developed using local data for specific jurisdictions, typically consisted of fewer than a dozen factors (e.g., the California Base Expectancy Tables developed in the 1960s), and focused on identifying groups of offenders with distinctly different risks of future offending. Many of today's instruments, often referred to as generation 3 or generation 4, have expanded beyond the singular objective of risk assessment to classify individuals by risk of delinquency. These instruments often contain dozens of factors (for example, the Correctional Offender Management Profiling and Alternative Sanctions [COMPAS] Youth risk assessment instrument). They frequently divide risk factors into two groups: "static" and "dynamic" (see, for example, Schwalbe, 2008; Hoge, 2002). Static factors are generally measures of prior delinquency. Dynamic factors are commonly referred to as "criminogenic needs" and represent conditions or circumstances that can improve over time (Andrews, Bonta, & Wormith, 2006). In addition, protective factors and references to "responsivity" have been added to generation 4 instruments. Responsivity is intended to reflect an individual's readiness for change and gauge a youth's ability to respond to particular treatment methods and programs (Andrews, 1990). Generation 4 instruments contain anywhere from 42 to approximately 150 factors. These variations in methodology and philosophy raised questions about which types of instruments most accurately and effectively help jurisdictions differentiate between low-, moderate-, and high-risk youth. Many evaluations of risk assessments based validity on correlation coefficients or other measures of association. Those that examined the degree of difference in recidivism rates observed for youth identified as low, moderate, or high risk often found little differentiation; results could vary substantially by race, ethnicity, and gender. Few jurisdictions conducted local validation studies to ensure a risk assessment's validity and reliability, and now one foundation-funded reform effort is telling agencies that local validation is not required if an instrument has been validated in three agencies or for similar populations. Perhaps the most significant change in the last few decades has been the emergence of commercially available risk assessment systems. Prior to this development, risk assessment studies were generally conducted by universities, nonprofit research organizations, or research units within government agencies. Claims made about the validity and reliability of some of these tools have been challenged by other researchers (Skeem & Eno Louden, 2007; Baird, 2009). In response to concerns about the classification and predictive validity of several risk assessments voiced by juvenile justice practitioners and researchers, OJJDP funded a proposal submitted by the National Council on Crime and Delinquency (NCCD) to evaluate commonly used risk assessments by comparing their predictive validity, reliability, equity, and cost. NCCD is a nonprofit social research organization, and its researchers conducted the study of eight risk assessments in 10 jurisdictions in consultation with an advisory board of juvenile justice researchers and developers of commercial juvenile justice risk assessment systems included in the study. Details: Oakland, CA(?): National Council on Crime and Delinquency, 2013. 541p. Source: Internet Resource: Accessed April 22, 2014 at: https://www.ncjrs.gov/pdffiles1/ojjdp/grants/244477.pdf Year: 2013 Country: United States URL: https://www.ncjrs.gov/pdffiles1/ojjdp/grants/244477.pdf Shelf Number: 132109 Keywords: ClassificationJuvenile Justice SystemsPredictionRecidivismReoffendingRisk Assessment Instruments |
Author: Drake, Elizabeth K. Title: Predicting Criminal Recidivism: A Systematic Review of Offender Risk Assessments in Washington State Summary: Under Washington State's sentencing laws, an adult convicted of a felony in superior court receives a sentence as prescribed within the ranges of the state's sentencing guidelines. Depending on the seriousness of the crime and a person's criminal history, some sentences may result in confinement in prison, community supervision, or both. The Department of Corrections (DOC) has jurisdiction over offenders sentenced to more than one year of confinement as well as those who receive a sentence of supervision in the community. In 1999, the Legislature enacted the Offender Accountability Act (OAA) that set state policy regarding the intensity of community supervision. The law requires DOC to classify offenders according to their future risk for re-offense and the harm they have caused society in the past. DOC must deploy more staff and rehabilitative resources to higher-risk offenders. Since the passage of the OAA, DOC has implemented two different risk assessments to assist with the classification of offenders. The 2009 Legislature required DOC to use a risk assessment "recommended to the department by the Washington State Institute for Public Policy as having the highest degree of predictive accuracy for assessing an offender's risk of re-offense." We focus our systematic review on assessments that have been tested on offender populations in Washington State. The Washington State Institute for Public Policy (WSIPP) was approached in 2012 by DOC to determine if a new risk assessment under consideration by DOC has the highest degree of predictive accuracy of future recidivism. To fulfill this legislative requirement, WSIPP systematically reviewed the literature on risk assessments that have been statistically "validated." That is, we examined tools developed and tested on offenders in Washington to determine the degree of accuracy of predicting recidivism. Details: Olympia: Washington State Institute for Public Policy, 2014. 4p. Source: Internet Resource: Accessed August 11, 2014 at: http://www.wsipp.wa.gov/ReportFile/1554/Wsipp_Predicting-Criminal-Recidivism-A-Systematic-Review-of-Offender-Risk-Assessments-in-Washington-State_Final-Report.pdf Year: 2014 Country: United States URL: http://www.wsipp.wa.gov/ReportFile/1554/Wsipp_Predicting-Criminal-Recidivism-A-Systematic-Review-of-Offender-Risk-Assessments-in-Washington-State_Final-Report.pdf Shelf Number: 132995 Keywords: PredictionRecidivismRisk Assessment |
Author: Early, Kristin Parsons Title: Validity and Reliability of the Florida PACT Risk and Needs Assessment Instrument: A Three-Phase Evaluation Summary: The Florida Department of Juvenile Justice (Department) began efforts in 2005 to develop a comprehensive, evidence-based system of assessing the risks and needs of youth referred to the juvenile justice system. A system change of this magnitude was not easily accomplished and required strong collaboration within the Department, as well as with juvenile justice stakeholders and community partners. The Department followed a long-range plan for developing and implementing its new risk and needs assessment instrument referred to as the Positive Achievement Change Tool (PACT). This process included pilot testing of the assessment and a Pre-Validation Study to norm the instrument to Florida's delinquency population and examine its initial validity in predicting offender risk to re-offend. The current evaluation examined the validity and reliability of the PACT in three phases: Phase I assessed the validity of the PACT risk and needs assessment in accurately predicting recidivism; Phase II involved confirmatory and exploratory factor analyses of all PACT assessment data to assess the utility and parsimony of PACT scoring; and Phase III examined consistency in PACT scoring through assessment of inter-rater reliability. The Justice Research Center (JRC) performed the analyses reported here under contract (Contract P2085) with the Department following a competitive procurement process. Details: Tallahassee: Justice Research Center, 2012. 125p. Source: Internet Resource: Accessed August 13, 2014 at: http://www.djj.state.fl.us/docs/probation-policy-memos/jrc-comprehensive-pact-validity-and-reliability-study-report-2012.pdf?Status=Master&sfvrsn=2 Year: 2012 Country: United States URL: http://www.djj.state.fl.us/docs/probation-policy-memos/jrc-comprehensive-pact-validity-and-reliability-study-report-2012.pdf?Status=Master&sfvrsn=2 Shelf Number: 133042 Keywords: Evidence-Based PracticesJuvenile Offenders (Florida)PredictionRecidivismRisk Assessment |
Author: Berk, Richard Title: Statistical Procedures for Forecasting Criminal Behavior: A Comparative Assessment Summary: There is a substantial and powerful literature in statistics and computer science clearly demonstrating that modern machine learning procedures can forecast more accurately than conventional parametric statistical models such as logistic regression. Yet, several recent studies have claimed that for criminal justice applications, forecasting accuracy is about the same. In this paper, we address the apparent contradiction. Forecasting accuracy will depend on the complexity of the decision boundary. When that boundary is simple, most forecasting tools will have similar accuracy. When that boundary is complex, procedures such as machine learning that proceed adaptively from the data will improve forecasting accuracy, sometimes dramatically. Machine learning has other benefits as well, and effective software is readily available. Details: Philadelphia: University of Pennsylvania, Department of Statistics and Department of Criminology, 2013. 36p. Source: Internet Resource: Accessed June 3, 2015 at: http://www-stat.wharton.upenn.edu/~berkr/Bake-Off%20copy.pdf Year: 2013 Country: United States URL: http://www-stat.wharton.upenn.edu/~berkr/Bake-Off%20copy.pdf Shelf Number: 135865 Keywords: Crime Forecasting Criminal Justice Policy Prediction |
Author: Moore, Robin, ed. Title: A compendium of research and analysis on the Offender Assessment System (OASys): 2009-2013 Summary: The compendium presents the OASys studies completed between 2009 and 2013, including a systematic review of the underlying evidence base, a survey of assessors' views and experiences, and analyses of various aspects of reliability and validity. Updated versions of the operational risk of reoffending predictors are presented. The findings support the continuing development of offender assessment. Details: London: National Offender Management Service, 2015. 367p. Source: Internet Resource: Analytical Series: Accessed July 30, 2015 at: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/449357/research-analysis-offender-assessment-system.pdf Year: 2015 Country: United Kingdom URL: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/449357/research-analysis-offender-assessment-system.pdf Shelf Number: 136266 Keywords: OffendersPredictionRecidivismReoffendingRisk Assessment |
Author: Spence, Douglas H. Title: Recidivism by Direct Sentence Clients Released from Day Report Centers in 2011: Predictors and Patterns over Time Summary: This study investigates the factors that predict the likelihood that DRC clients will be arrested, booked into jail, or incarcerated within 2 years of release. It also examines the timing of recidivism events during the period after release. The strong relationship between successful program completion, risk scores, and recidivism provides evidence of the impact of DRC programming and the predictive validity of the LS/CMI risk assessment tool. Analysis of LS/CMI subcomponent scores reveals important areas of criminogenic need for the DRC client population in WV, and suggest means for further improving the quality of service delivery in DRCs. Findings related to the timing of recidivism point to additional opportunities for reducing recidivism rates through the use of targeted post-release supervision strategies. Implications for quality assurance, effective treatment dosage, and adherence to evidence-based practices are also discussed. Details: Charleston, WV: Criminal Justice Statistical Analysis Center, Office of Research and Strategic Planning, 2016. 27p. Source: Internet Resource: Accessed February 3, 2016 at: http://jrsa.org/sac-spotlight/wv-recidivism/wv-drc-recidivism.pdf Year: 2016 Country: United States URL: http://jrsa.org/sac-spotlight/wv-recidivism/wv-drc-recidivism.pdf Shelf Number: 137750 Keywords: Alternative to IncarcerationsDay ReportingDay Reporting CentersOffender Risk AssessmentPredictionRecidivismRisk Assessment |
Author: Davidson, Leighann J. Title: Evidence-Based Offender Assessment: A Comparative Analysis of West Virginia and U.S. Risk Scores Summary: This study describes the results of more than 8,000 LS/CMI risk assessments provided to West Virginia offenders in 2013 and 2014. West Virginia normative data is compared to U.S. offender population norms derived from assessment data gathered from nine states across the country. Results indicate West Virginia has a lower risk offender population compared to other states - this is true regardless of correctional setting (i.e., community or institutional confinement). Approximately 74-76% of West Virginia offenders under correctional supervision have risk scores that are below the U.S. average. The low risk population under supervision, in part, explains the comparatively low recidivism rates observed in West Virginia. Compared to other states, West Virginia offenders have lower levels of need in most areas, especially the Procriminal Attitude/Orientation and Antisocial Pattern domains. The study results suggest there may be substantive differences in the risk and needs of male and female offenders in West Virginia. Consideration of LS/CMI risk scores may enhance the state's efforts to manage its correctional population, protect the public, and save resources. Details: Charleston, WV: West Virginia Criminal Justice Statistical Analysis Center, Office of Research and Strategic Planning, 2015. 14p. Source: Internet Resource: Accessed February 3, 2016 at: http://www.djcs.wv.gov/ORSP/SAC/Documents/Davidson%20et%20al%202015_EBP%20Offender%20Assessment%20Comparative%20Analysis.pdf Year: 2015 Country: United States URL: http://www.djcs.wv.gov/ORSP/SAC/Documents/Davidson%20et%20al%202015_EBP%20Offender%20Assessment%20Comparative%20Analysis.pdf Shelf Number: 137752 Keywords: Offender Risk Assessment Prediction Recidivism Risk Assessment |
Author: Spence, Douglas H. Title: The Predictive Utility of Risk and Needs Assessment Summary: Risk and needs assessment plays a crucial role in determining the services offenders receive while in correctional custody and their level of supervision after release. According to the principles of effective correctional intervention, clients assessed as having a higher risk of recidivism should receive both a greater treatment dosage and a higher level of case supervision. This strategy of providing more services to higher risk individuals is frequently described as adhering to the "risk principle" (Andrews and Dowden, 2006). In order to adhere to the risk principle, however, correctional programs must first ensure that they are accurately assessing offenders' risk and needs. The Level of Service/Case Management Inventory (LS/CMI), and its predecessor the Level of Service Inventory-Revised (LSI-R), are two of the most prominent and widely-used tools for assessing offenders. Both have been subjected to extensive empirical research and have been shown to accurately predict the likelihood of recidivism for a variety of offender populations (Vose, Cullen and Smith, 2008). The LS/CMI is currently used by all correctional agencies in West Virginia to assess risk for recidivism. The tool is completed through a process that involves an offender interview combined with the use of official records. The collective information is used to calculate risk scores that indicate an overall risk for recidivism as well as identify specific criminogenic needs (i.e., dynamic risk factors shown to be empirically related to recidivism). These factors include: education/employment, family/marital relationships, substance abuse, procriminal attitudes, antisocial peers, leisure/recreation activities, antisocial personality, and past criminal behavior. LS/CMI scores are utilized to make a variety of decisions including level of supervision and services to be provided to protect public safety. Several recent and forthcoming studies conducted by researchers from the Office of Research and Strategic Planning (ORSP) assess the effectiveness of the LS/CMI for predicting recidivism by offenders in WV. These studies investigate the statistical relationships between various offender characteristics (including LS/CMI scores) and the likelihood of committing a new offenses during a 24 month follow-up period. Details: Charleston, WV: Criminal Justice Statistical Analysis Center, Office of Research and Strategic Planning, 2015. 5p. Source: Internet Resource: Research Brief; Evidence-Based Practice Series, No. 1: Accessed February 3, 2016 at: http://www.djcs.wv.gov/ORSP/SAC/Documents/JCEBP%20Research%20Brief%201_final.pdf Year: 2015 Country: United States URL: http://www.djcs.wv.gov/ORSP/SAC/Documents/JCEBP%20Research%20Brief%201_final.pdf Shelf Number: 137753 Keywords: Alternative to IncarcerationsDay ReportingDay Reporting CentersOffender Risk AssessmentPredictionRecidivismRisk Assessment |
Author: Stavrou, Efty Title: The revised Group Risk Assessment Model (GRAM 2): Assessing risk of reoffending among adults given non-custodial sanctions Summary: Aim: To re-examine the Group Risk Assessment Model (GRAM) for predicting reoffending in adults given non-custodial sentences and to assess the accuracy of the model. Method: Adult offenders given non-custodial sentences in 2011 were the cohort of interest. Reoffending within 24 months of the index appearance was measured using court data. Models predicting reoffending using personal, index offence and criminal history characteristics were undertaken using multivariate logistic regression and model fits were assessed. Model validity and reliability was also measured by applying the model estimates to sub-group data and to separate smaller cohorts. Results: Of the 81,199 adult offenders, 26% reoffended within two years of the index appearance. The best model fit for GRAM 2 comprised age, gender, Indigenous status, number of concurrent offences, prior custodial sentence, prior proven offences and the index offence type. The internal and external validity of the model was strong, however application of the model to offenders from smaller geographical areas or to those with a prior history of prison or property offending should be undertaken with care. Application of the model for screening purposes should also be carefully considered. Conclusion: The GRAM 2 has been shown to be a robust tool for predicting reoffending. Although reliable, model estimates and their applicability should be re-examined periodically. Details: Sydney: NSW Bureau of Crime Statistics and Research, 2016. Source: Internet Resource: Contemporary Issues in Crime and Justice, No. 197: Accessed October 12, 2016 at: http://www.bocsar.nsw.gov.au/Documents/CJB/Report-2016-GRAM2-Group-Risk-Assessment-Model-CJB197.pdf Year: 2016 Country: Australia URL: http://www.bocsar.nsw.gov.au/Documents/CJB/Report-2016-GRAM2-Group-Risk-Assessment-Model-CJB197.pdf Shelf Number: 145439 Keywords: PredictionRecidivismReoffendingRisk Assessment |
Author: LeCroy & Milligan Associates, Inc. Title: Assessing Risk of Recidivism Among Juvenile Offenders: The Recidivism Risk Instrument. Technical Report Summary: The past four decades have been witness to an increasing interest in risk assessment in the corrections field. Risk assessment is based on the calculation of statistical relationships between offender characteristics and outcomes such as recidivism. The process of risk assessment involves estimating an individual's likelihood of continued involvement in delinquent behavior, based on the relationship of specific characteristics to delinquency (Gottfredson & Moriarty, 2006; Krysik & LeCroy, 2002). Several trends have contributed to the increased popularity of risk assessment. A steady increase in the number of juveniles that were entering the juvenile justice system has heightened the demand for rehabilitation services. This increased demand for services combined with their high cost has prompted efforts to target services, based on a systematic assessment of need, to those at the high end of the risk continuum, while reducing efforts aimed at those on the low end. The assignment of low risk cases to intensive services may not only be a waste of scarce resources, but may in fact be criminogenic (Andrews et al. 1986). Statistical risk assessment is increasingly being used to replace assessments based on "clinical" judgments which are subjective and less accurate than statistical instruments. Actuarial/statistical risk instruments generally classify youth as low-, medium-, or high-risk for recidivism by estimating an offender's likelihood of reoffending based on their similarity to others who have recidivated in the past. Accordingly, the goal of statistical risk instruments is to identify a group of offenders with different rates of recidivism and focus intensive treatment interventions on those offenders with the greatest risk of returning to custody. Research has shown that a small number of offenders contribute disproportionately to the crime rate. For instance, research on two cohorts of first-time juvenile delinquents in Orange County, California found that approximately 10% of the juveniles accounted for over one-half of all subsequent offenses (Kurz & Moore, 1993). Based on these findings, Orange County developed a risk-based intervention strategy that emphasizes risk rather than crime seriousness. The recognition that a relatively few individuals commit the majority of crimes has prompted a more streamlined approach to the early identification of the most persistent juvenile offenders. The purpose of identifying high-risk juveniles early in their criminal careers is to provide them with cost effective prevention and treatment services. In Orange County, the chronic offender population averages nearly 20 months of incarceration within 6 years of their first offense, making the cost of incarceration alone $44,000 per individual in 1993 dollars (Kurz & Moore, 1993). At the rate of approximately 500 new chronic juvenile offenders per year, the estimated cost for incarceration in Orange County is $22 million per cohort. A reduction in placement would result in significant cost savings. Further, there is reason to suspect that predictors of recidivism for boys differ from predictors of recidivism for girls (Emeka & Sorensen, 2009). For example, Plattner and colleagues (2009) identified sex specific predictors of recidivism among a sample of incarcerated youth. For boys, the strongest predictors for recidivism were age at first incarceration and presence of oppositional defiant disorder. For girls, the strongest predictors for recidivism were dysthymia (protective factor) and generalized anxiety disorder. Consistent with previous work, early aggressive or disruptive behavior was not a good predictor of later delinquency for girls. The most common problem encountered in risk prediction research is data limitations. Data limitations constrain the potential for sophisticated and more appropriate statistical approaches to analysis. There are two basic sampling issues that lead to limitations in the data. First, the size of the sample is critical. In terms of how big the sample should be, Jones (1996) recommends at least 500, half for estimation and half for validation. If a large number of variables are being tested in multivariate statistical analysis, it is common practice to ensure that the sample includes at least 10 subjects for each predictor variable considered (Norman & Streiner, 1986). Second, the sample must be representative of the population to whom the instrument will be applied; therefore, it should be a random sample. Even if a sample is large and appropriately drawn, serious problems may still emerge. The patterns found in one sample can lead to overestimating patterns that might exist in other samples. Representativeness can encompass the variables of age, gender, race and ethnicity, regional area, and time period (Jones, 1996). Criticism of several studies has revolved around the use of only one sample for estimation, and the subsequent failure to test the accuracy of the derived model on an independent validation sample (Krysik & LeCroy, 2002; Schwalbe, 2007). The primary purpose of using a separate sample for validation is to test the extent that empirically derived relationships persist across samples. When the risk assessment instrument is validated on the same sample from which it was estimated, the rate of correct classification is naturally much higher. Thus, the use of at least two samples is recommended, one for estimation and one or more for validation. The lack of differentiation on the criterion variable is always more apparent during validation than the construction of the instrument. The prediction instrument developed on a selective sample is often applied to a population containing a wider range of risk than that of those individuals originally studied. Under such circumstances, the best policy is to identify a random sample that is as closely related as possible to the population of interest. If this is not possible, it may be useful to examine empirically differences between the original sample and the population of interest. Invariably the best laid plans are constrained by the quality of the data available. Often this problem is not recognized, or it may be noticed and not addressed. The main effect of missing data is to reduce the size of the sample at the stage of multivariate analysis. How this problem is dealt with depends in part on how much data is missing, and how important the particular variables afflicted are thought to be as predictors. If there are few missing values and the data are missing completely at random, then the analysis should be based on those cases with a complete set of variable values (Jones, 1996). Other than a reduced sample size, this complete case approach poses no problems. An alternative approach that makes use of available information is to include all cases that have values for a specified group of variables. This available-cases approach has the significant disadvantage that statistics such as means and variances are based on samples of different sizes. A third approach is the imputation of missing values. This involves the estimation of missing values based on those data that are available (Little & Rubin, 1987). In instances where a person's risk-level is assessed at more than one point in time, it is necessary to move away from a reliance on variables that remain constant toward more dynamic indicators. Static indicators can be historical (e.g., parent criminality) or ascribed (e.g., gender or race). As individuals can exercise no control over static factors, they are insensitive to change over time. The repeated use of these same variables can result in individuals being censured over and over for the same attributes. Psychiatric measures, response to supervision or institutionalization, employment, and family situation, are examples of dynamic factors. One risk assessment instrument involving dynamic factors is offered by Baird (1984). He has developed an initial risk assessment instrument and a reassessment instrument. His reassessment instrument retains the most significant initial predictors such as age at first adjudication, prior criminal behavior, and institutional placements of more than 30 days, and adds to this dynamic factors such as response to supervision and the use of community resources. Dynamic factors introduce a stronger element of judgment or discretion into the classification process. Underwood (1979) cautions that the inclusion of subjectively scored items may provide opportunity for personal biases to be passed off as scientific judgment. The goal in risk assessment is to choose the smallest number of variables with the greatest predictive validity. This goal, however, can be modified by the issue of face validity. Burnham (1990) argues that decision makers feel uncomfortable with only a limited set of data items and require a range of information, most of which they do not take into account. He differentiates between information, that which leads to predictive efficacy; and noise, those items necessary for the instrument to be supported by the user. Most commonly, prediction models include both individual and environmental variables as predictors. Ideally, the pool of possible predictors is theoretically derived, with one variable representing each theoretical construct, and each of the selected variables tested for validity and reliability. In practice, prediction in the area of criminality is constrained by poorly defined theory. Given these cautions, we turn our attention toward key predictor variables supported in the literature. Details: Tucson, AZ: LeCroy and Milligan Associates, 2012. 35p. Source: Internet Resource: Accessed October 12, 2016 at: http://www.lecroymilligan.com/data/resources/recidivism-risk-instrument-final-report-10312012-final-revision-b.pdf Year: 2012 Country: United States URL: http://www.lecroymilligan.com/data/resources/recidivism-risk-instrument-final-report-10312012-final-revision-b.pdf Shelf Number: 140672 Keywords: Juvenile OffendersPredictionRecidivismRisk Assessment |
Author: LeCroy & Milligan Associates, Inc. Title: Arizona Department of Juvenile Corrections Risk Assessment Project Findings Summary: The primary purpose of this project was to construct an interim risk tool for the Arizona Department of Juvenile Corrections (ADJC). The function of the interim risk tool is to statistically estimate the likelihood that an offender will continue to be involved in delinquent activity, and classify the offender according to their relative risk of continued involvement (Gottfredson, 1987; Krysik & LeCroy, 2002). The goal of statistical risk assessment is to effectively group offenders by risk level in order to allocate resources for higher-risk youth while maintaining validity and consistency in the assessment and decision-making process. The development of an interim risk tool was constructed through an integrated three step approach: (1) a review of the literature related to modeling risk in juvenile correctional populations was conducted; (2) the interim risk tool was developed and subsequently validated on a separate, independent sample; and (3) risk tool administrators and users were surveyed to determine their perceptions and use of the current ADJC risk prediction instrument. Historically, the type of information included in risk instruments includes the offender's criminal history, social history (e.g., substance abuse, education/employment, family background, psychological profile), and demographics (i.e., age, gender, race/ethnicity) (Cottle et al., 2001; Krysik & LeCroy, 2002). In most risk tools, the likelihood of recidivism is most closely related to a few consistent variables - criminal history, substance abuse, family background, and school performance. Details: Tucson, AZ: LeCroy & Milligan Associates, 2006. 32p. Source: Internet Resource: Accessed October 12, 2016 at: http://www.lecroymilligan.com/data/resources/adjcriskassessmentprojectfindingsreportfinal-1.pdf Year: 2006 Country: United States URL: http://www.lecroymilligan.com/data/resources/adjcriskassessmentprojectfindingsreportfinal-1.pdf Shelf Number: 140673 Keywords: Juvenile CorrectionsJuvenile OffendersPredictionRisk Assessment |
Author: Thornton, Sara Title: Predicting Serious Domestic Assaults and Murder in the Thames Valley Summary: Thames Valley Police uses a risk assessment model to identify those cases of domestic violence where the risk of future harm is high. This study looked at all the cases on serious domestic assault and murder between 2007 and 2009 to establish how accurate the risk assessments had been in predicting the serious harm. In 55% of cases there was no prior recorded contact with the police. In only five out of 118 cases was the case assessed as high risk. Effectively there was an 80% false negative rate. In the same period 1740 other victims were assessed as high risk arguably resulting in a 99% false positive rate. A case control study was carried out to try to identify any risk factors that marked out those offenders who committed the most serious domestic assaults from other violent offenders. The case control study found that those who committed serious domestic assault and murder were less criminogenic than the risk pool of all violent offenders - contrary to the central hypothesis of escalating violence. The study also found that male offenders who committed serious domestic assaults were more than three times likely to be suicidal than other violent offenders. Details: Cambridge, UK: Wolfson College, University of Cambridge, 2011. 98p. Source: Internet Resource: Thesis: Accessed October 17, 2016 at: http://www.crim.cam.ac.uk/alumni/theses/Thornton,%20S.pdf Year: 2011 Country: United Kingdom URL: http://www.crim.cam.ac.uk/alumni/theses/Thornton,%20S.pdf Shelf Number: 145094 Keywords: Domestic ViolenceHomicideIntimate Partner ViolencePredictionRisk AssessmentViolence Against Women |