The Erroneous Use of and Reliance on the LSI-R

The Use And Abuse Of The LSI-R In Parole Evaluations Challenging So-Called “Objective” Testing

By Eric Marcy, Esq.

The Level of Service Inventory-Revised “LSI-R” is frequently misused in the context of parole evaluations. Providing an air of scientific reliability, its application in the forensic setting is actually unreliable, subjective, and highly dependent on factors that may be decades old, failing to take into consideration, relevant facts relating to maturity and rehabilitation for which it is not designed to account. While there is authority that supports the efficacy of the LSI-R for determining security risk, classification, programming and custodial placement at the time of initial incarceration, the LSI-R was never designed for and is not appropriate to predict an inmate’s prognosis on parole after years of incarceration.

An internet search will disclose a variety of articles seeming to validate the predictive reliability of the LSI-R, but many of these articles suffer from flaws in the sampling techniques used, including generalizing testing across race, sex, and age. The true test of its efficacy for determining prediction of recidivism would require studies involving similarly situated sample inmate populations. Moreover, some “raters” throw around “recidivism” “statistics” without disclosure (or even understanding) the underlying sampling of the source of such “statistics.” Without the disclosure of data sample and sampling techniques, the efficacy of the purported data sample upon which the “statistics” are generated is misleading. Such statistics are meaningless if the sample population upon which the percentages are generated do not represent a similar population to the individual being evaluated. If the underlying data base is not representative of individual being evaluated, the “statistics” are not only meaningless but are, in fact, misleading.

A review of the LSI-R reveals that it is essentially a two page true/false check and 0-3 rating checklist that purportedly determines risk based upon a factors such as, criminal history, education, financial, substance abuse, and “attitudes” and then grades out a score risk level. The fallacy (when offering measures of risk for release on parole) of this test is that virtually all of the factors relate to and are determined by an inmate’s social, economic, family, and addiction situation that existed at the time of initial incarceration. The LSI-R does not account for rehabilitation programming, work performance, or other positive information that may be relevant over years of incarceration that would be a better indicator of the ability of an inmate to adjust and succeed on parole.

It becomes unreliable and misleading when it is administered to an individual who has been incarcerated for many years. Individuals who have been incarcerated for decades may have committed a crime when very young, may have matured, successfully participated in numerous rehabilitative programs, obtained educational degrees while incarcerated, and been a model inmate. The problem is that the factors used to score the test are frozen at the time of initial incarceration. The test itself is probably most accurate for determining an initial risk assessment, classification and custodial admission planning, not for assessing suitability for release on parole. It is not a predictor of the ability of a subject to comply with conditions of parole.

The LSI-R does not evaluate a subject for success in residential Supportive Community Programs, such as halfway house type programs, or the suitability for and likelihood for success in a supportive non-residential community based program. Given the nature of the test, if the same psychologist administered these same tests ten or even twenty years from the initial evaluation, the same, or close to the same, results would likely occur. The factors being used to score the test are “static” in that they address the subject’s situation immediately prior to incarceration. Those static factors will not change over time.

Test training manuals caution that the LSI-R should only be used as a component of a thorough clinical evaluation. The training manuals contain specific sections for “Notes/Special Circumstances” and a “Professional Discretion Over-ride Section.” The training manual itself states that in approximately 10% of cases the override section changes the score. How an evaluator determines an over-ride is a subjective assessment and will greatly vary depending on who is conducting the evaluation. Remarkably and frequently the professional administering the test gives no consideration to important relevant factors. There are factors that the test does not account for and that should be considered to adjust the score. In discussing the LSI-R, one manual, the Department of Corrective Services, LSI-R Training Manual, specifically notes the test’s limitations and cautions about its use as a forensic tool in the wrong context or in the improper administration. Frequently, the test is administered based upon an incomplete record of the inmate’s history and/or based upon hearsay information that has not been verified for accuracy.

The training manual specifically identifies the limitations of the LSI-R, confirming that is only to be used to assist in a professional evaluation, does not determine a level of service, and may be inaccurate as to some types of crimes. The manual cautions that the LSI-R is not an assessment of mitigating and/or aggravating factors for use at sentencing and not intended to determine a sentence.

The proper use and consideration of the over-ride section is critical and requires the evaluator to consider all relevant facts. The training manual notes that the evaluator must ask the question “[d]oes this result fit?” The test itself involves a very subjective evaluation by the evaluator, to determine “[i]f the result seems to be wrong for the individual offender.” Such a test cannot remotely be considered “objective.” Its use must be scrutinized in the context of the inmate’s entire life, not just the circumstances that existed at the time of the crime and events prior to incarceration. If the test is to be used honestly and ethically, intervening events that are not captured by the instrument must be considered and accounted for as part of the over-ride section. A mechanical application of the checklist without an assessment of an inmate’s full and complete record will not provide an accurate assessment and will be misleading. A proper use of the override would discuss and address additional intervening rehabilitative programming, work, and training the subject participated in as well as accurate current assessments as to the subject conduct and insight while incarcerated.

The use of the LSI-R in determining suitability for parole is highly suspect and should be scrutinized and challenged, especially if the factors generated to score the test involve factual circumstances that date back years or even decades. If the LSI-R is to be used at all, it should be combined with an honest clinical assessment of the subject’s full institutional history, including all positive as well as negative information and all psychological and classification evaluations performed during the course of years. Most importantly, if it is to be used there should be an honest and fair use of the override section. Simply put, the LSI-R cannot be characterized as an “objective” psychological test and purported “statistics” regarding “recidivism” should not be advanced under the guise of scientific authority.

Our criminal defense and parole lawyers at Lyons & Associates, P.C. will vigorously defend you as you go through the criminal justice and parole process. Call us at 908-575-9777 or contact us online today to schedule a free consultation. We have offices located in Somerville and Morristown, New Jersey, and represent clients throughout New Jersey including, but not limited to, clients throughout Somerset County, Morris County, Middlesex County, Union County, Essex County, Bergen County, Hunterdon County, Warren County, Sussex County, Passaic County, and Monmouth County.