Updating the Guidelines of the Sex Offender Risk Assessment Instrument

REPORT ON LEGISLATION BY THE SEX OFFENSE WORKING GROUP, CRIMINAL COURTS COMMITTEE, CRIMINAL JUSTICE OPERATIONS COMMITTEE AND CORRECTIONS AND COMMUNITY REENTRY COMMITTEE

S.3096 (Sen. Krueger)

AN ACT to amend the correction law, in relation to risk assessment instruments for sex offenders

THIS BILL IS APPROVED

I.  INTRODUCTION

This report is respectfully submitted by the Criminal Courts Committee, the Criminal Justice Operations Committee, and the Corrections and Community Reentry Committee (the “Committees”) of the New York City Bar Association.[1]  The City Bar is an organization of approximately 24,000 members dedicated to improving the administration of justice. The Committees include members from a broad spectrum within the criminal justice field.

The Committees have thoroughly analyzed the legal, social, and fiscal implications of the Bill.  The Bill amends New York’s Sex Offender Registration Act (“SORA”) to require the Board of Examiners of Sex Offenders (the “Board”) to consult a “validated risk instrument” when it makes its recommendation to courts regarding the appropriate risk level to assign a person convicted of sex-related offenses.  The Bill requires the Board to validate the instrument through “a periodic retroactive study at least every five years to determine the predictive value of the risk assessment instrument.”  After each study, the Bill mandates that the Board submit a written report to the Governor and the Legislature explaining its findings and recommending changes to enhance the predictive capabilities of the instrument.  The Bill requires the Board to complete the first study within two years of the Bill’s effective date.

The Committees urge passage of the Bill. Although it has been used to classify tens of thousands of individuals, the current Risk Assessment Instrument (“RAI”) employed by the Board is not a valid method to predict these individuals’ likelihood of reoffending.  Even when it was created in 1996, the RAI was based on outdated research.[2]   In the 25 years since its creation, the Board has never substantively updated the RAI, even though most advances in the field of predicting recidivism rates for individuals convicted of sex-related offenses have occurred during those years.  Furthermore, the RAI lacks a coherent methodology and seems to have been designed arbitrarily.  Finally, studies show that many factors used by the RAI to predict an individual’s risk of reoffending do not correlate with risk, and that there are more predictive factors that are not included in the instrument.

There is no justification for the Board’s continued reliance on the RAI.  Several validated risk assessment instruments exist and are used by other states.  The Board’s reliance on an inferior instrument harms not only registrants, but also the public.  SORA is only effective if individuals are accurately assessed so that the public and law enforcement can direct their limited resources toward monitoring the most high-risk individuals.  Currently, the Board’s use of the RAI dilutes the effectiveness of SORA by both under- and over-estimating individuals’ risk.  We urge that the Legislature pass the Bill in order to improve New York’s implementation of SORA.

II.  OVERVIEW OF SORA

Individuals in New York who have been convicted of a “sex offense” are required to register as sex offenders.[3]  Sex offenses include, among others offenses, rape (including “statutory rape”), sexual abuse, possession of child pornography, and kidnapping where the victim was under 17 and the offender was not the victim’s parent.[4]  They also include what are termed “sexually motivated felonies,” crimes that would not otherwise be considered sex offenses had they not been committed for the offender’s “direct sexual gratification.”[5] At risk assessment hearings conducted by a trial court, individuals are classified as Level One, Level Two, or Level Three.

These classifications affect individuals’ registration requirements as well as the extent to which they are subjected to community notification.  Level One registrants, unless they are subject to certain statutory designations,[6] must register with the Division of Criminal Justice Services (“DCJS”) for 20 years, while people classified as Level Two and Level Three must register for life.[7]  Additionally, Level One and Level Two designees must register their address by mail annually and in person every three years, while Level Three designees must register in person every 90 days.[8]  Detailed information about Level Two and Level Three registrants including the nature of their offense, whether their offense involved pornography or a computer, their photograph, a description of their physical appearance, their home address, and their place of employment is available on New York State’s official online Sex Offender Registry.

There are many additional restrictions on registrants that are not contained in SORA itself.  For instance, federal law prohibits individuals subject to lifetime registration requirements (i.e. Level Two and Level Three registrants in New York) from receiving federal housing subsidies such as Section 8, or living in federally funded housing such as the New York City Housing Authority.[9]  This restriction affects not only the individual, but any household which includes an individual with a lifetime registration requirement, thus affecting entire families.  Moreover, State law prohibits Level Three registrants and anyone whose victim was under the age of 18 years who are on parole or post-release supervision from “entering” within 1,000 feet of a school.[10]  In New York City, which has over 1,800 public schools and countless private schools, this restriction makes it nearly impossible for individuals to find suitable housing.[11]

At SORA hearings, trial courts designate individuals’ risk levels.  In most cases, the Board provides the individual, the court, and the prosecution with a Case Summary as well as a completed RAI prior to the hearing (the Board does not weigh in on cases where the individual is sentenced to probation and adjudicated at sentencing). Although the courts have the final say regarding an individual’s ultimate designation, in practice, the RAI score usually dictates the outcome of a SORA hearing.  Courts are instructed that departures from the risk level calculated on the RAI are to be “the exception, not the rule.”[12]

III.  THE RAI’S SIGNIFICANT FLAWS

The RAI is based on outdated research, has never been validated by New York State despite its frequent use and its significant implications on both registrants and public safety, and has been criticized by numerous experts who specialize in diagnosing and treating individuals convicted of sex-related offenses.

The Board created the RAI in 1995, with assistance from the Colorado Division of Criminal Justice (CDCJ).  After reviewing CDCJ’s suggested guidelines, the Board invited professionals involved in criminal justice and law enforcement to review the RAI.[13]  Over a two-day period, these professionals applied the RAI to a mere 20 cases, making what they perceived as appropriate modifications.[14]  In January 1996, the Board adopted the modified RAI still in use today.[15]

The RAI contains 15 items within four categories: current offense(s), criminal history, post-offense behavior, and release environment.  Various weights are given to each category.  An individual is assessed points based on whether he meets the criteria in each of the categories.  The RAI generates a score ranging from 0 to 300, which corresponds to a level designation.  A score of 70 points or below translates into a presumptive Level One; a score of 75 to 105 into a presumptive Level Two; and a score of 110 to 300 into a presumptive Level Three.  As mentioned previously, people classified as Level One must register for 20 years.  But the RAI also provides “overrides” if the individual meets one of four characteristics: they have a prior felony conviction for a sex crime; they inflicted serious physical injury or caused death; they made a recent threat that they will reoffend by committing a sexual or violent crime; or a clinical assessment of the individual shows that they have a medical abnormality that decreases their ability to control sexual behavior.  These individuals will be designated as “sexual predators,” “sexually violent offenders,” or “predicate sex offenders,” respectively, and required to register for life.[16]

The RAI relies on outdated research: the most recent study relied on to create the RAI was published in 1995, and some of the studies were even older.[17]  The instrument has never been revised or updated.[18]  The RAI’s reliance on outdated research is especially troubling because the most significant scientific advancements in the field of predicting recidivism rates for individuals convicted of sex-related offenses have occurred in the years since the RAI’s creation.[19]

The RAI’s risk factors are intended to take into account a variety of concerns, including an individual’s risk of reoffending, the harm that would be caused by reoffending, and the difficulty law enforcement and prosecutorial agencies may have in detecting and convicting those who reoffend.[20] This hodgepodge covers and conflates too many disparate concerns with no scientific basis for doing so. As Justice Daniel Conviser of the New York State Supreme Court, New York County, observed:

 [T]he instrument mixes and matches purportedly objective factors related to the risk of re-offense, with numerical value judgments about the degree of harm an offender’s conduct causes and policy considerations.  Some RAI factors are based on “harm,” some on recidivism risk, some on both of those factors, some on policy grounds and for some factors, it is not clear on what basis points are assessed.[21]

In other words, the RAI lacks a coherent methodology.

In addition, the range of points assessed for each RAI factor varies greatly, and the Board has provided no reason for this seeming arbitrariness.  The range under Factor 1, for instance, is 10 to 30 points, while the range under Factor 14 is 0 to 15 points.  Rationality is also lacking within each factor.  Factor 9, which addresses an individual’s criminal history, provides just one example.  The RAI assesses three times more points under Factor 9 if an individual has a prior non-violent felony conviction than if they have a non-violent misdemeanor.  However, although an individual’s criminal history may be relevant to their risk of recidivism, the differences between specific felony and misdemeanor convictions can be negligible (compare, for example, Sexual Misconduct, Penal Law § 130.20(1), a misdemeanor, with Rape in the Third Degree, Penal Law § 130.25 a felony), and there is no research to support this dramatic disparity in the number of points.[22]

Even worse, many of the factors that the Board claims measure an individual’s likelihood of recidivating do not appear to predict recidivism at all.[23]  These include Factor 2 (sexual contact with the victim), Factor 3 (number of victims), Factor 4 (duration of offense conduct with victim), Factor 6 (other victim characteristics), Factor 8 (age at first sex crime), Factor 10 (recency of prior felony or sex crime); Factor 12 (acceptance of responsibility), and Factor 14 (supervision).[24]

In child pornography cases, which make up about 10% of cases that trigger SORA registration, the RAI routinely overstates the risk of re-offending, resulting in predictably inaccurate risk level adjudications. The Board itself seven years ago published a Position Paper acknowledging that the RAI produces an “unintended, anomalous result as the majority of offenders convicted of child pornography cases will be scored the same when there are clearly vast differences amongst these types of offenders[.]”[25] The Position Statement clarified that points should not be scored in child pornography cases under Factor 3 (number of victims) and Factor 7 (relationship with victim).

While the Board generally follows the Position Statement by not scoring in these categories but instead recommending upward departure when there are very large numbers of images or other factors that increase the risk of re-offense or threat to public safety, the courts read Factors 3 and 7 mechanically, increasing points in these categories in virtually every case, resulting in a Level Two adjudication for all people convicted of possessing child pornography who fail to articulate grounds for downward departure.[26] Replacing the RAI with an empirically tested assessment instrument would resolve this anomaly, and would require that courts increase risk levels only where there is a rational basis to do so.

The RAI’s treatment of an individual’s age provides another illustration of its flaws, and particularly its reliance on outdated research.  The RAI assesses 10 points under Factor 8 if the individual was 20 years old or younger at the time of their first sex crime (assessing no points if the individual was older than 20).  But the studies and articles relied on by the Board to support its conclusion that someone’s young age increases their risk of recidivism were, even in 1996, “decades old.”[27]  Nonetheless, more than two decades later, “the Board uncritically relies on these articles and applies their findings, apparently universally, to any individual whose first sex offense was committed before the age of 20 to reach its conclusion of a high risk of re-offense.”[28]  Contrary to what the RAI suggests, however, more recent and reliable empirical research demonstrates that sexual recidivism rates reported for individuals who commit sex offenses as juveniles are generally lower than those reported for individuals who commit sex offenses as adults.[29]

Factor 3 serves as another example of a factor that lacks any scientific validity.  Under this factor, an individual is assessed 20 points if the sex offense involved two victims and 30 points if the offense involved three or more victims.  Although studies show that the number of past sexual offenses committed by an individual is relevant to recidivism, current research does not support the association between the number of victims and an increased risk of recidivism.[30]  Indeed, an individual’s commission of many sex offenses against the same victim may be more indicative of their risk of reoffending than the commission of one offense against two different people.[31]

Additionally, the RAI’s use of four “overrides” that require individuals, no matter their point score on the instrument and risk level adjudication, to register for life, has also been criticized.  Experts believe that no one factor should be used to assign a risk level or an override.[32]  Doing so is also contrary to SORA itself, which directs the Board to make recommendations after examination of all relevant factors.[33]

The RAI’s flaws are reflected in practical reality:  it frequently overestimates individuals’ risk, at great cost to those individuals, their families, law enforcement and society as a whole.[34] About 25% of New York registrants – 10,756 people – are classified as Level Three, indicating a high risk of recidivism.[35]  But studies suggest that the percentage of people who present a high reoffending risk is actually considerably lower.  A study conducted by the U.S. Department of Justice – before most states had established sex offender registries – found that only 5.3% of former offenders are likely to be rearrested for a sex crime during the first three years after their release.[36]  Similarly, a meta-analysis of 82 recidivism studies found that just 13.7% of nearly 30,000 former offenders were arrested or convicted for a new sexual offense five to six years following their release.[37]  And a study from 2008 found that first-time offenders commit over 95% of New York’s sex crimes.[38]  Thus, New York’s classification of almost one-quarter of registrants as “high risk” cannot and does not accurately reflect reality and, as discussed further below, diverts state resources from supervision of individuals who do, in fact, pose a high risk of reoffending.  Furthermore, while evidence suggests that “the RAI may tend to inaccurately increase more risk level classifications than it actually decreases . . . the arbitrary character of the instrument means that many [individuals] will also be classified as being at a lower risk to re-offend than is justified.”[39]

Even though it has now been used to classify 52,945 registered people in New York,[40] the State has never validated the RAI.[41]  Nor is there currently any provision in SORA requiring that the RAI be validated.[42]  However, in 2011, Dr. Jeffrey C. Sandler of the University of Albany conducted a study of 3,663 registered sex offenders in New York within five years of their release to determine the predictive value of the RAI.[43]  He found that an individual’s score on the RAI marginally corresponds to their risk of re-offending, but that a better-designed risk assessment instrument would much more accurately predict an individual’s risk.  Dr. Sandler further concluded that the RAI does not significantly predict the harm caused by an individual’s reoffending – the second aim of the RAI and a goal mandated by the Legislature.[44]  The RAI’s failure to achieve its second goal of predicting future harm is not surprising.  Its attempt to do so is based on a flawed premise that when an individual reoffends, he will commit the same type of offense.[45]

Because it has not been updated in more than 25 years, the RAI also fails to consider factors that have since been found to affect recidivism.  Factors that militate against recidivism include an individual’s time spent in the community offense-free, their age at the time of release, and whether they have lived with an intimate partner for two years.  By contrast, factors that predict future offending include the existence of female or extrafamilial victims, “stranger” victims (victims who were known to the individual less than 24 hours) and the presence of “paraphilic” interests in the individual being assessed (i.e., pedophilia, exhibitionism, voyeurism or other paraphilic disorders).[46] The RAI considers none of these factors.

Not surprisingly, in light of the RAI’s numerous flaws, two groups of experts who regularly treat individuals convicted of sex-related offenses strongly support passage of the Bill.  The New York State Alliance for the Prevention of Sexual Abuse and the New York State Association for the Treatment of Sexual Abusers have issued a joint position statement criticizing the Board’s continued use of the RAI and recommending use of validated instruments instead.[47]

IV.  THE AVAILABILITY OF VALIDATED RISK ASSESSMENT INSTRUMENTS

The Board’s sole reliance on the RAI is particularly troubling in light of the availability of validated risk assessment instruments.  As of 2008, at least 25 states were using validated risk assessment instruments to determine how much supervision to give registrants.[48]  The most widely used instrument is the Static-99R, but some states use the Minnesota Sex Offender Screening Tool and the Vermont Assessment of Sex Offender Risk. A separate instrument known as the Child Pornography Risk Tool (“CPORT”) based on empirical data acquired from individuals arrested for charges involving child pornography was published in 2015[49] and has started to be widely used to assess risk of re-offense for sexual crimes in individuals arrested for charges involving child pornography. Using a separate set of criteria like the CPORT in hands-off child pornography cases could help resolve the systematic scoring errors created by using the RAI in such cases. Thus, although the Bill leaves the choice of which validated instrument to use in the hands of the Board, many options are available, each of which will almost undoubtedly produce more accurate results than does the RAI.

V.  THE COSTS OF MISCLASSIFICATION

The RAI’s deficiencies have a detrimental impact not only on individuals adjudicated under the instrument and their families, but on the public as well.  With the Board’s use of a validated risk assessment, the public will be better informed about individuals who may pose a high risk of recidivism, and low-risk individuals will be able to better reintegrate into society without the harsh stigma, publicity and burdensome restrictions that accompany high risk classifications.

a.  Counterproductive Effects on the Registrant

The current RAI has a profoundly detrimental impact on individuals convicted of a sex offense and their families, and may even lead to increased recidivism.

First, the RAI is inappropriately and unfairly punitive in that many of the factors considered under the RAI do not relate to an individual’s likelihood of reoffending.  For example, despite having “no current empirical basis for being correlated with an increased risk for sexual offense recidivism,”[50] one factor in the RAI is whether the sexual contact was over or under clothing.  Notably, the Penal Law does not make any such distinction:  sex abuse is charged the same whether contact is over or under clothing.[51]  Another factor is acceptance of responsibility, which has no shown correlation either to the likelihood of re-offense or to the harm likely to result from any re-offense.

Second, because of the RAI’s deficiencies, a large number of people are wrongly burdened by the restrictions imposed on and stigma associated with Level Two and Level Three classification.  These include the legally mandated restrictions on federal housing assistance for registrants and their families discussed above.  In addition, public notification makes it extremely difficult for individuals to obtain private housing, and obtain and maintain employment.  Level Two and Three registrants are required to report the name and address of their employer, if they have one, and the address is posted on the SORA online registry.  Not surprisingly, businesses may be reluctant to appear to be employing people on the registry, and this posting may therefore act as a deterrent to employment.  In one study of 30 registrants, 83% reported that they had been excluded from a residence and 57% reported that they had lost employment as a result of their status as a sex offender.[52]

Studies have found that these restrictions actually may increase the likelihood of recidivism.[53]  When burdened with these restrictions and the hazards they create, people who under normal circumstances would be unlikely to reoffend may be more likely to engage in antisocial behavior.[54]  This results directly from excessive restrictions and pervasive social stigmatization that accompany a sex offense conviction.  In fact, experts have concluded that the most effective way to prevent recidivism is to allow an individual to reintegrate into society.[55]  And even homelessness has been negatively affected, since certain shelters that once housed large numbers of individuals subject to residency restrictions – including many individuals referred by their parole officers upon release from prison – are now off limits to those individuals because the shelters are deemed to be too close to schools.[56] The result is that some individuals have been moved to remote areas, far from family and resources, and that other individuals have remained in prison long past their ordinary release dates for lack of suitable housing options. [57]

In addition to the burdens imposed by higher risk classifications, individuals wrongly designated as Level Two or Three registrants bear stigma and, in some cases, harassment and physical abuse.[58]  As noted by Professor Roger Lancaster, sex offender registration laws “were promoted as applying to the ‘worst of the worst,’ . . . [but] [i]n practice, they turn expansive classes of people into pariah outcasts who can never be reintegrated back into society.”[59]  This is particularly true in New York where, because of the arbitrary nature of the RAI, numerous individuals who pose little or no threat to public safety are being wrongly classified and treated as the “worst of the worst.”

b.  The Detrimental Effect on the Public

The RAI interferes with one of the primary purposes of SORA: to protect New Yorkers by informing them of registrants in their community who pose the highest risk of reoffending.[60] As discussed above, many of the factors considered under the RAI bear no correlation to the likelihood of recidivism.  Because the RAI does not accurately determine risk, the SORA registry is not a reliable source of information, and citizens have no satisfactory means of obtaining information about individuals who may pose a threat to the community.

As a result of the RAI’s over-inclusiveness, the sheer number of individuals listed on New York’s online sex offender registry is so large that the registry is not a meaningful source of information.  A total of 26,887individuals are registered as Level Two or Level Three sex offenders on New York’s registry.[61]  While this vast number of individuals listed as moderate and high risk on the registry effectively generates public fear and outcry, it does little to actually inform the public of safety risks.  As Justice Conviser noted, “[a] mother who learns to her horror that a ‘high risk’ sex offender has moved into her neighborhood has no way of knowing that the instrument used to make this designation has [rarely] been tested to see if it is accurate and is based on information which was outdated more than a decade ago.”[62]

In addition to providing the public with more accurate information about high risk individuals, a validated risk assessment instrument will serve the second fundamental purpose of SORA: “enhancing law enforcement authorities’ ability to fight sex crimes.”[63]  Currently, because law enforcement is required to monitor individuals who are wrongly classified as high risk, resources are diverted from other enforcement priorities.  Indeed, a 2009-10 report from the New York Senate Standing Committee on Crime Victims, Crime, and Correction noted that “there are far too many people in New York who are misclassified in the higher levels of risk, and therefore unnecessarily diverting limited resources from likely re-offenders.”[64]  At the same time, “inaccurate classifications allow high-risk offenders who are wrongfully characterized as being at a low risk to re-offend to escape enhanced scrutiny and thereby present a greater threat to public safety.”[65]

With a more accurate risk assessment instrument, resources can be more effectively used to monitor those who may actually pose a threat to public safety and remaining resources can be used for other purposes such as treatment programs and sexual abuse prevention initiatives.  As noted by psychologist Dr. Laurie Guidry, “if the resources necessary to protect the community from high-risk sexually dangerous predators must be applied to . . . all [individuals] who are rated Level 3 [under the current RAI], our current resources are not adequate.  On the other hand, if the most dangerous predators can be differentiated from those who pose far lower risk, the community safety resources can be targeted to those who require the greatest degree of supervision and surveillance.”[66]

VI.  APPLICABILTY OF A VALIDATED RISK ASSESSMENT INSTRUMENT TO PREVIOUSLY-CLASSIFIED INDIVIDUALS

The Committees are aware that some members of the Legislature may be concerned that enactment of the Bill may lead to an influx of previously-registered individuals requesting new SORA hearings.  The Committees urge the Legislature to pass the Bill despite this concern.

Both the New York State Court of Appeals and the Second Circuit have held that the purpose of SORA is to promote public safety.[67]  Yet at this juncture SORA – because of its reliance on the RAI – effectively serves to punish large numbers of individuals who do not truly pose a risk. Because validated risk assessment instruments more accurately identify individuals at higher risk of sexually reoffending, public safety would be better served if the Bill were applied retroactively, giving previously-registered individuals the right to a new risk assessment hearing using the most accurate, up-to-date, and sound risk assessment instrument available.  Individuals who present a high risk of reoffending can thus be more reliably identified, and individuals who do not present such a high risk can be re-adjudicated to more appropriately identify their lower risk of reoffending.  In turn, resources will be expended more efficiently by focusing on individuals that truly need heightened community supervision.

The proposed legislation does not present a radical program, but instead one that is consistent with the purpose of the registration law and with legislative intent. Indeed, the New York State Legislature has already recognized that an individual’s risk of reoffending may change over time and the Correction Law currently provides that the individual may apply to the Court annually to modify the individual’s risk level.[68]  In this regard, the public’s interest in accurate findings of risk is consistent with the individual’s due process right to reliable assessments of risk.[69]

Finally, retroactive application of a validated risk assessment instrument will not create significant costs.  In the first place, as previously noted, SORA currently allows both Level Two and Level Three registrants, annually, to petition sentencing courts for a modification of their risk level.[70] Since this is the same population that might apply for a new SORA level determination using a validated risk assessment instrument, it appears that retroactivity would not significantly increase the current burden on the judicial system.  Proper classification of individuals may even reduce the State’s overall expenditures by making it easier for them to access housing, employment and other services and thus reducing their reliance on the State.

Any increased expenditure of judicial resources will be offset by applying the correct level of supervision of people on the registry.  As discussed above, the RAI greatly over-predicts the risk of reoffending.[71]  Adjudication under a validated risk assessment instrument will place more individuals at their appropriate low or moderate risk to reoffend, saving vast community-supervision resources.

VII.  CONCLUSION

As Justice Conviser aptly observed in McFarland: “If sex offender risk assessment is worth doing (and the Legislature has clearly stated that it is), it is worth doing correctly.  The stakes, to our cherished liberties and to our families and children are too high. . . . [W]e can and must do better.”[72]

By enacting the Bill, New York will better protect all of its citizens, including people on the sex offender registry as well as the community at large.  In short, by enacting this legislation, New York will do better.

Reissued February 2022