The Level of Service Inventory-Revised (LSI-R) is a risk assessment instrument designed to identify adult offenders’ likelihood of reoffending and is widely used in corrections. The instrument has been built on extensive empirical evidence about offender risk factors and pilot-tested extensively prior to its formal release. The LSI-R is part of a new, third generation of offender risk instruments known as risk/need assessments, because its items include a balance of historical and demographic static risk factors and an array of dynamic criminogenic need factors. This article describes the LSI-R, its place among LSI instruments, its content and widespread use, its training protocol, and its validity.
Background of the LSI-R
The LSI-R is an adaptation of Donald Andrews’s original LSI Level of Supervision Inventory, (LSI-VI). In fact, the LSI was administered and researched for more than a decade before its release as the LSI-R. By the time the LSI-R was published in 1995, there were at least 25 research papers on the LSI-VI that established its psychometric properties, including its inter-rater and test–retest reliability, internal consistency, construct validity (i.e., the extent that it correlated with other measures of constructs such as propensity for rule violation), and its predictive validity (i.e., its correlation with subsequent recidivism and other antisocial behavior such as rule violation). However, with the exception of a few studies, mostly with Canadian federal offenders, this research was conducted on offenders who were on probation or in provincial facilities in Ontario, Canada. Regardless, the LSI-VI served, in essence, as a pilot instrument affording many opportunities for risk/need assessment research prior to the formal release of the LSI-R by Andrews and his colleague, James Bonta, in 1995.
Content and Organization of the LSI-R
The LSI-R is very similar to the LSI-VI in content. It includes the same 54 items grouped into the same 10 domains or sections: Criminal History (10), Education/Employment (10), Financial (2), Family/Marital (3), Accommodation (3), Leisure/Recreation (2), Companions (5), Alcohol/Drug Problems (9), Emotional/Personal (5), and Attitudes/Orientation (4). (The LSI-VI test option to include write-in items listing specific conditions for probation or parole has been eliminated.) The LSI-R is available in a paper-and-pencil format on a carbon paper Quick-Score Form that allows for expedient manual scoring. The publisher, Multi-Health Systems, also offers a Windows-based computer program that generates a profile report, a comparative report, and a group report.
Norms were developed for male and female offenders, listing the percentile ranking of offenders throughout the range of possible scores on the instrument. In addition, five risk levels were created for male inmates with labels and 1-year recidivism rates (return to custody) as follows: high (72%), medium/high (57.3%), moderate (48.1%), low/moderate (31.1%), and low (11.7%). The authors also suggested guidelines for the classification of offenders in various contexts including guidelines for the amount of supervision of probationers (minimum, medium, and maximum); placement to a halfway house (appropriate, appropriate with close supervision, and inappropriate); and inmate classification to security level, with separate cutoff scores for recommendations for male and female prisoners.
An easily overlooked but important symbolic difference between the LSI-IV and LSI-R was its change in name. With the development of the original LSI, the objective was to determine the level of supervision that should be afforded to each probationer. But that was seen as limiting since, as a dynamic risk/need assessment tool, the LSI-R offered direction for a full range of services to offenders by identifying an offender’s particular criminogenic needs. Moreover, as research was developing on the importance of having a full range of services to offer offenders in the community, the function of probation officers was evolving from traditional supervision to direct service provision and brokering or advocating for services to address offenders’ specific criminogenic needs. Hence the original term, level of supervision, was replaced by the more inclusive level of service terminology.
Rise in the Popularity of the LSI-R
The commercial release of the LSI-R brought the tool to criminal justice agencies (primarily corrections), academics, and clinicians worldwide. Growing use of the LSI-R occurred both geographically and across multiple correctional contexts. This included prisons where it was introduced to predict recidivism, to select offenders for release, and to classify offenders according to security level. Some agencies also began to conduct LSI-Rs as part of the presentence investigation. Although the instrument was never designed to offer advice to the court about the severity of the sentence, it can be used to inform conditions of the sentence, including orders for treatment or programming or behavioral conditions (e.g., the offender would be instructed not to consume alcohol or to associate with certain individuals).
Because of its growing popularity in large correctional agencies, some with tens of thousands of offenders to assess per year, and its lengthy time to complete, often a number of hours for an offender interview, a thorough review of various documents, and collateral checks, a short version of the LSI-R, the LSI-R: Screening Version, was released in 1998. It consists of 8 items and can be used to determine which offenders are most likely to be high risk and therefore deserve a complete LSI-R assessment.
The decade following the publication of the LSI-R saw a dramatic increase in its use in the field and in research pertaining primarily to assessing its predictive validity with respect to recidivism. By 2010, LSI assessors logged more than 1 million administrations of the LSI-R, its youth version/ Case Management Inventory (YLS/CMI), and its successor (LS/CMI), in a single year. But, this popularity also led to new concerns. Any risk assessment tool, regardless of its inherent value, is only as good as the assessor who administers it.
LSI-R Training
In order to accommodate the LSI-R’s growth and popularity and at the same time protect its integrity, standard training packages were developed and a system of user certification was created. A “train the trainer” approach was introduced to address the sizable, widespread demand for training from North America, Europe, and more recently, Pacific Rim countries.
To begin, the authors built a training curriculum that included theory and research on criminal behavior, detailed instructions on administration of the tool, practice exercises that increased in complexity, and testing of both the content and the administration of the instrument. Typically, trainers are handpicked members of a correctional organization, who demonstrate a deep understanding of and appreciation for the LSI-R and its underpinnings, as well as having other kinds of social skills that enable them to work effectively with their colleagues. Upon successful completion of training, trainers are certified by the publisher as LSI-R trainers and mandated to offer high-quality training to field staff.
A mechanism was established to create master trainers, individuals who were authorized to train trainers to give LSI-R training. Most master trainers are experts in the field, typically having a combination of research experience, professional training, expertise in the field of corrections, and experience administering LSI-Rs and conducting LSI-R training. Master trainers also include criminal justice professionals who do LSI user and trainer training as part of their own practice or consulting service. With the proliferation of the LSI-R, it quickly became apparent that having master trainers was essential to meet the need for trainers in large correctional agencies, some of which have hundreds of probation officers.
Research on the LSI-R
Upon its publication, the LSI-R generated a great deal of offender risk/need research by academics and correctional agencies in Canada, the United States, and beyond. Although it was succeeded by a subsequent version (LS/CMI) in 2004, in 2013, it was the most widely researched version of the LSI, with more than 50 predictive validity studies.
These studies examined the capacity of the instrument to predict antisocial behavior. General recidivism was the most common outcome criterion but others included violent recidivism and failure on supervision in various community contexts including probation, parole, and community residential settings or halfway houses.
The LSI-R research firmly established the instrument’s predictive validity. It did so with various offender groups, including women and ethnic minorities, in various jurisdictions, particularly Canada and the United States, in custody and community. However, there remains some debate about the equivalence of the LSI-R across gender, race, ethnicity, and jurisdiction, particularly countries. Andrews’s initial position was that risk factors, including criminal history, criminal attitudes, criminal associates, and lack of opportunity through education and employment, are universal risk factors that merit consideration in the assessment of any potential offender. Some critiques have suggested that these factors might not be equally applicable across all demographic groups and national borders but have differing levels of importance or weighting, while others have suggested that factors related to social location, including gender, race, and poverty, lead to other delinquent pathways that are not captured in the LSI-R. This could lead to bias in the administration and use of LSI-R. Although some individual studies have shown disparities along these dimensions, Mark Olver’s comprehensive meta-analysis of the Level of Service Scales revealed consistent predictive validities of total score by race and gender but not by nationality (i.e., lower in United States) and ethnicity (i.e., lower among Hispanics).
As a third generation risk assessment tool, one of the research objectives of the LSI-R was to illustrate its dynamic predictive validity. This was achieved by demonstrating the improvement of latter reassessments of offenders in terms of their statistical relationship with recidivism. A popular way of examining LSI-R reassessments was to compare recidivism rates of offenders who went either from low risk during the initial assessment to high risk upon reassessment or vice versa, from high risk to low risk. To be clear, changes of this magnitude in the 8- to 12-month intervals used in these studies are dramatic, the former being a recipe for disaster and the latter an indication of considerable optimism. Follow-up analyses bore out these prognostications. In five studies, offenders who went from high risk to low risk had, on average, a 37% lower recidivism rate than those who remained at high risk. Conversely, those who went from low risk to high risk had, on average, a 26% higher recidivism rate than those who stayed at low risk.
With the emergence of other offender risk assessment tools, there was a noteworthy sparring match in the early 2000s between the LSI and Robert Hare’s Psychopathy Checklist–Revised. The Psychopathy Checklist–Revised had emerged at about the same time as the LSI-R, but it was formally designed as a measure of offender psychopathy and not as a risk assessment instrument, although it had been applied as such both clinically and in research. Currently, the consensus is that there is little, if any, difference between these two instruments, and a number of other scales, in terms of statistically derived measures of predictive validity. However, they do have numerous other differences, such as their theoretical orientation and applicability in a clinical and programmatic purpose. The current wisdom is that these other characteristics should guide the choice of risk assessment instruments.
Final Thoughts
The LSI-R was crafted as a risk/need offender assessment instrument both to predict offender recidivism and to guide planning and programming with the offender. There is no doubt that it accomplishes the first goal. The extent to which it has achieved its second goal is less clear. Studies in a number of jurisdictions have demonstrated that the tool is not used to its full capacity when it comes to case planning and intervention. In fact, case management and intervention may be based more heavily on the court’s recommendations or conditions than the criminogenic needs identified on an LSI-R. This is not surprising for two reasons: one, the authority wielded, justifiably, by the judge; two, the fact that there is nothing inherent in the LSI-R that explicitly guides or directs case planning and intervention. This led to another (fourth) generation of offender risk assessment instruments, one that was characterized by the LS/ CMI. Nonetheless, the LSI-R remains one of the most widely used offender risk/need assessment tools throughout the world.
References:
- Andrews, D. A., & Bonta, J. L. (1995). The Level of Service Inventory–revised: User’s manual. Toronto, Canada: Multi-Health Systems.
- Andrews, D. A., & Bonta, J. L. (1998). The Level of Service Inventory–revised: Screening Version. User’s manual. Toronto, Canada: Multi-Health Systems.
- Bonta, J., & Andrews, D. A. (2017). The psychology of criminal conduct (6th ed.). New York, NY: Routledge.
- Gendreau, P., Goggin, C., & Smith, P. (2003). Is the PCL-R really the “unparalleled” measure of offender risk? Criminal Justice and Behavior, 29, 397–426.
- Hannah-Moffat, K. (2009). Gridlock or mutability: Reconsidering “gender” and risk assessment. Criminology & Public Policy, 8, 209–219.
- Olver, M. E., Stockdale, K. C., & Wormith, J. S. (2014). Thirty years of research on the Level of Service scales: A meta-analytic examination of predictive accuracy and sources of variability. Psychological Assessment, 26, 156–176.
- Wormith, J. S. (2011). The legacy of D. A. Andrews in the field of criminal justice: How theory and research can change policy and practice. International Journal of Forensic Mental Health, 10, 78–82.