to "Update" List
Click to Notify Us
In Court Today
Bills in the
Letters / e-mails
How To Access
No SO Zones
--- How the Mental Health System calculates "Recidivism Statistics" ---
The Mental Health System (MHS) compiles "recidivism" statistics completely different from the CJS, and for different purposes. The MHS is focused on "prediction of a mental condition, or a relapse" and "treatment of that condition." That is the focus of the MHS.
Recidivism in the MHS has a different meaning than in the CSJ system. Here it means, a mental condition is, likely or not likely, to re-occur. In CJS it means commission of another crime. Yes, a subtle and very important difference to community concerns.
Before we delve into "Predicting Recidivism with Statistical Tools," here are some past studies which cite "sex offender recidivism statistics" using different statistical tools. -1- -3- -4- -5-. Notice the varied use of terms, child molester, rapists, incest offenders. What these are talking about is, "SUBGROUPS of SEX OFFENDERS," not the class of offenders.
Notice how the quoted "recidivism rates" keep changing, by sub-groups, so frequently that you don't know the true recidivism rate for all sex offenders. This is the essence of how the MHS expresses Recidivism Rates by Sub-Groups of offenders.
Many folks don't realize, that a Pedophile (see DSM-IV pgs 527-8) is a sub-group of sex offenders, as are Ephebophiles. The distinction being, the age of the victim (13 & under [prepubescent], AND THOSE, 14 to 17 [adolescent]). Confusion regarding definition of pedophile occurs, and is caused by persons who quote that term to mean all sex offenders, and is part of the reason for this research. Sub-Groups of Sex Offenders
"Any discussion of the causes of the sexual abuse of children should distinguish between pedophila and ephebophilia. A pedophile, as mentioned in the previous chapter, is an adult who has recurrent, intense, sexual urges and sexually arousing fantasies involving a prepubescent child. The age of the child is arbitrarily set at 13 years or younger and the adult is at least 5 years older than the child. An ephebophile is an adult who has recurrent, intense sexual urges and sexually arousing fantasies but whose object is a pubescent child or adolescent. The age of the child is arbitrarily set at 14 through 17 years. Again, the adult is at least 5 years older than is the child.
Knowing these distinctions is critical to understanding what you hear or read.
A distinction made by Groth, Hobson, and Gary (1982) is significant. They divided pedophiles into two groups, regressed and fixated. Regressed pedophiles were described as individuals with a primary sexual orientation toward adults of the opposite sex. Under conditions of extreme stress, the regressed pedophile may psychologically regress to an earlier psychosexual age and engage in sex with children. The typical case is a male whose wife is emotionally unavailable to him and who thus turns to his daughter for sex. Fixated pedophiles/ephebophiles have a primary sexual interest in children or teens; they rarely, if ever, engage in sex with peers. In most cases, they are described as exclusively interested in children or teenagers."
NOTE:So that our focus is not misunderstood our intent is not to discredit these reports, but to explain them in a way that, helps folks to fully understand, rather than understand in a hysterical sense. Many folks, including legislators and judges, don't understand what they are hearing and seeing. Then the result is misapplied in their decisions.
___ Misunderstanding ___
We ask professionals to bear with our layman, and very general, descriptions, as we are trying to explain concepts to non-professionals, and any clarification assistance is appreciated.
___ The Process of Creating Statistical Tools ___
Researchers generally follow this, or a similar, process: First they decide what it is they want the assessment tool to predict. ex: violence. Earlier studies, if any, are considered and blended into this new prediction tool.
Next a group of offenders are chosen, those who have already displayed the characteristics of the study, often from prison. Their backgrounds analyzed, using proven assessment tools, and like or similar characteristics are identified and recorded.
Validating Assessment Tools: A mandatory step to prove if the tool does what it is supposed to do. i.e., predict violence (our example). Applying its theory to a group of offenders (w/like or similar characteristics) to see if it predicts, i.e. violence (our example), then determine its correlation factor.
Now, the culling process begins, to arrive at a initial base group who most closely display the characteristic the researcher is trying to predict. Of those offenders who meet the criteria, a random selection of them decides who will be in the final "Purified Base Group (PBG)!"
Following release, the PBG will be tracked for a period of time, often for several years, with contacts at specific intervals, to record what happens related to the goal of the study, and to review their files at that time. Often this can only be done if the offender is returned to prison.
The characteristics of the PBG become the characteristics, which later offenders will be compared to -to see how closely s/he correlates-. A close correlation will result in a rating for the new offender that says, s/he is likely to be violent (our example). This system becomes a "Assessment Tool" which tests for violence in offenders. Oversimplified, yes, but the concept is our point.
Note: Some tools are developed by combining other validated tools into one new tool. (no base group of offenders, instead the new tool is validated by testing it against other groups of offenders to see if it results in a high correlation factor.)
An effective tool has a good correlation factor, meaning it is able to predict the "violence (our example)" significantly more times than it fails to detect the violence. However, for our discussion we are not going to focus on the validation phase and assume that is being done, and correctly by an independent source.
"Assessment Tools Available Today:" We will use their acronyms, HCR-20, IM-RAT, J-SOAP, J-RAT, LSI-R, LSI-SV, MMPI-2, MnSOST-R, PCL-R, PCL-SV, RRAS, RRASOR, SACJ-Min, SAI, SAI-J, SCR-20, SHQ-R, SIR, SONAR, SORAG, SRA, SRA-99, SVR-20, Static-99, VASOR, VRAG, VRS, VRS-SO, and many others. We have compiled a list of these risk assessment tools with further information about them.
Statistical Tool Foundations, Inherent Design Errors
__and, Applying Tools to the Wrong Offenders__
The Perfect Tool: Statistical tools have been created to predict various different offender characteristics (violence, domestic violence, juvenile recidivism, adult recidivism, etc.) with the ultimate goal of predicting "recidivism." Our tools listing, shows over 25 different tools, that alone says there is no one perfect tool, researchers are still striving to perfect a single recidivism tool that works for every time and for every offender.
Revised tools: Notice several of the tools indicate, revised version (LSI-R, MnSOST-R, PCL-R, SHQ-R, etc.), pointing out shortcomings or problems with the initial version, some of which are still in use. Further, there are some (SORAG) which are an improved version of a different tool (VRAG). The history of any tool is critical to know.
Purified base groups (PBG): Earlier we described how a PBG is chosen, and effectively it is a group of like offenders, each having very close characteristics. This raises certain issues:
Combinations of tools: The Static-99 (Hanson & Thorton, 1999) is one such tool being a combination of the Rapid Risk Assessment for Sex Offense Recidivism (RRASOR - Hanson, 1997)) AND the Structured Anchored Clinical Judgement tool (SACJ-Min, Grubin 1998). Reportedly this did improve the reliability a small amount to .71. The point is, throughout the United States there is a trend to combine tools (often by the States themselves) and there is little or no period of testing before use.
Currently even more tools are being developed, others being revised, and new combinations being tried, it is an evolving field of research, and each offender assessed becomes an unwilling part of the overall experiment!
Rating System Used: Today's risk assessment systems, when an offender is rated he is always a degree of risk (low-medium-high). The systems are inherently skewed due to using "Purified Base Groups," every offender in the base group is a degree of risk.
Age, Gender and Disabilities: Look closer at our "Tools Listing" and notice, certain tools are used for "ADULT MALES," and others for "JUVENILES," and some for "WOMEN." Indicating age and gender should be factors, but are they factors in the tools?
A risk assessment system is designed to evaluate, not in the sense of fairness, but incremental from "0" to some "High." The absence of "0" means the system is designed biased. Further, when a system should start at "0," and instead starts at "1" it means everyone is rated at one level higher than they should be; biased and flawed!
Note: If you were to take the characteristics of an offender who was hospitalized, or who had passed away, they would still calculate to a degree of risk [low-medium-high]. SYSTEMS ARE INHERENTLY SKEWED!
Tools Used on Wrong Subgroups of Offenders:During our research we found, researchers commented this-tool or that-tool, should not be used on this/that type of offender (child molester, rapist, incest offender, etc). Apparently false ratings occurs with certain offender groups, and certain tools. Knowing this, you would think the very first thing on all coding forms would be "USE ONLY FOR ____ / DO NOT USE FOR ___," but such is absent in every system.
Base Group Recidivism Rates:When risk assessment systems are developed, researchers publish the recidivism rate for the base group, based on the period they tracked them. Once you realize how base groups are chosen you will realize its inherent recidivism rate would normally be high, because that PBG has like offenders in it, each having the characteristics of recidivism. However, this fact, is not published -along with the high base group recidivism rate-. Speakers often quote base group recidivism rates, without explanation, or simply ignore that fact, opting for -it fits my needs-.
"Age-as-a-Factor (age of offender)," age-at-conviction is used, but that is not the distinction between tools for "ADULTS" and "JUVENILES." Further, it is well know that, recidivism decreases as a person gets older. (Hanson 2001 [Summary] [Full Report-29 pgs]; and, 1-23-2003 Findlaw Report: Age/Disability) Age-as-a-factor (of the offender) is not included in any tool. WHY??
Crime type and degree: Our research revealed, no weight is given to crime type or degree that an offender is convicted of, yet, aggravating circumstances are considered and assigned weights. This heavily biases the system towards higher ratings with no counter weight.
"Age-as-a-Factor (age of offense)," the DSM-IV (pg 2 Severity & Course Specifiers) as appropriate. "A DSM-IV diagnosis is usually applied to the individual's current presentation and is not typically used to denote previous diagnoses from which the individual has recovered. The following specifiers ...: Mild; Moderate; Severe; In Partial Remission; In Full Remission, and Prior History.)
Disabilities, especially those resulting from illnesses (cancer, heart disease, etc), and due to age, should play a discretionary part in a tool. As folks age and/or reside in nursing homes, or mental health facilities (including commitment centers), need to be considered in the final analysis.
Does the absence of Age and Disabilities skew the tools making results more onerous? Eliminated because they mitigate???
The current purpose of risk assessment is to determine "likelihood of reoffense" [low-medium-high] which translates into level of community notification. It is generally thought that higher ratings mean, more concern about this offender. However, likelihood of reoffense does not translate to degree of concern!
If the truth be known, a community would not be as concerned over, a person who urinates in the park (or masturbates in a car)as a person who is likely to be a predatory rapist or murderer. Hence a "concern" factor is more important than a "recidivism" factor. The "concern" factor flows from the crime type committed by the offender.
The very heart of risk assessment systems is community concerns, therefore, systems cannot stop at "likelihood of reoffense," they must include a "degree of concern factor." The absence of this factor, paints all offenders equal, and establishes unwarranted fears in the community. Inclusion of crime type and degree BEGINS to accomplish this!
___ In Conclusion ___
___ All of the known risk assessment systems in use today: ___
Our research has identified inherent problems with all tools used today. Systems begin with a bias, systems are used on the wrong offender subgroups, there are no standards for expressing recidivism rates which confuse everyone, systems do not consider mitigating factors such as -age- -disability- -crime type or degree-, and systems have no community concern factor.
Sex offenders have been singled out based upon a false perception!
No wonder all offenders are considered the same, and it is no wonder that the community is hysterical about all offenders! The truth is being withheld from the community by the foregoing.
Given this is well known, why does Congress not act?