Evaluating New York State’s Sex Offender Management and Treatment Act

Abstract

The current study evaluated New York State’s sex offender civil management screening process. Specifically, the study sought to (a) determine which variables contributed significantly to offenders being deemed high risk and receiving a psychiatric examination, and (b) assess the likely public safety impact of civil management by examining the recidivism rates of a historical sample of offenders matched to Sex Offender Management and Treatment Act–reviewed offenders on variables important to the civil management screening outcome. Results of the screening analysis indicated the civil management risk assessment process places the most emphasis on variables empirically established to be related to sexual recidivism risk, such as the offender’s Static-99 score and whether the offender had a male victim. Results of the public safety impact analysis indicated that the review process appears to be accurately identifying the highest risk offenders, thereby likely increasing public safety.

Keywords

sex offenders civil management screening out evaluation risk assessment

In response to a growing perceived need to protect the public from sexual offenses, numerous public policies have been developed to manage and track convicted sex offenders. Although the most widely known of these initiatives are sex offender registration and community notification laws, many states have recently also enacted sex offender civil management laws (also commonly referred to as civil confinement/commitment laws, or sexually violent predator [SVP] laws). These civil management laws are designed to protect the public while offering treatment to the offender, by allowing for the confinement of those sex offenders deemed to be at high risk for sexual recidivism after they have completed their criminal justice sentences for certain sexual crimes. Twenty-two jurisdictions (20 states, District of Columbia, and the federal government) have enacted such laws.

On April 13, 2007, New York State (NYS) enacted its Sex Offender Management and Treatment Act (SOMTA, 2007) to civilly manage those sex offenders deemed to be at high risk for sexual recidivism. Specifically, SOMTA allows the state to civilly manage sex offenders about to be released from state supervision who have “mental abnormalities that predispose them to engage in repeated sex offenses” (p. 2), by either (a) placing the offenders in the community on strict and intensive supervision and treatment (SIST), or (b) civilly confining those offenders deemed to be dangerous (i.e., to be unable to control their sexual offending behavior).

Although the statute identifies which offenders are to be reviewed for possible civil management, what the milestones should be in the decision process, what information should be available to those evaluating the offenders, and that a finding of mental abnormality making the offender likely to sexually recidivate is ultimately necessary to recommend civil management, it does not explicitly state what criteria should be used to screen offenders out of the review process or to send them on to the next milestone. Thus, the SOMTA review process, which was designed by the case reviewers in line with evidence-based practices, is able to change to reflect new research findings as they emerge, and is in fact mandated by law to do so (SOMTA, 2007). As discussed below, the current study sought to inform the SOMTA review process and the sex offender civil management literature by examining the following: (a) which variables are most influential in the SOMTA review process, and (b) whether the process in NYS is correctly identifying high risk offenders (and, therefore, likely increasing public safety).

Civil Management in NYS

Screening process

According to SOMTA (2007), all offenders convicted of either a sexual felony or designated sexually-motivated felony who have a pending release from their sentences are referred to the Office of Mental Health (OMH) by whichever agency has jurisdiction over them (the Department of Correctional Services [DOCS], the Division of Parole, OMH, or the Office of Mental Retardation and Developmental Disabilities). OMH risk assessment and record review staff then conduct a brief review of each offender’s record and, if an offender has a qualifying offense and more than one known victim¹ in his or her record, refer that offender to either a multidisciplinary review (MDR) team or, depending on the offender’s score on the Static-99 (Hanson & Thornton, 2000),² a case review team (CRT) for a more thorough risk evaluation. If the offender reaches the CRT level of review (whether the result of MDR bypass or MDR referral) and the CRT deems the offender to be at high risk for sexual recidivism, the CRT then refers the offender for a psychiatric examination to determine if the offender has a diagnosable mental abnormality related to his or her sexual offending. If such a mental abnormality is found and the CRT finds the offender is an extremely high risk for sexual recidivism, the offender is then recommended for civil management and referred to the Office of the Attorney General (OAG) for legal processing. (All decisions made during the risk-review process are informed by the empirical sex offender recidivism literature, and a detailed overview of NYS’ civil management risk assessment and record review process can be found in Figure 1.)

Legal process

Once an offender has been recommended by OMH for civil management, the OAG then reviews the offender’s file and decides whether or not to move the case forward. If the OAG decides to pursue civil management, a bench hearing (i.e., no jury) is held to determine whether there exists probable cause to believe the respondent suffers from a mental abnormality related to sexual offending and is in need of civil management. On finding probable cause, a trial is scheduled to determine whether, by clear and convincing evidence, the respondent suffers from a mental abnormality as is statutorily required. If such a mental abnormality is found, the court then decides whether the respondent requires civil confinement in a secure treatment facility, or whether the respondent can be effectively managed in the community on SIST.

Evaluations of Civil Management Laws

Screening process

There have been three recent attempts to evaluate the process of screening offenders for possible civil management, all of which were conducted in Florida. The first two, Levenson (2004) and Levenson and Morin (2006), compared 229 offenders, who received a face-to-face evaluation (the second level of review) and were then designated for civil commitment, to 221 offenders who received a face-to-face evaluation and were not designated for civil commitment (i.e., were released). Although both studies found differences between the two groups in terms of offender, offense, and diagnostic characteristics, the contributions of these variables to the screening process (i.e., designating need for civil management) were confounded by the mental abnormality component necessary for civil commitment designations. That is, group membership for the participants was not based solely on evaluator assessments of sexual recidivism risk, but rather (and largely) on the offender being found to suffer from a mental abnormality related to volition to commit a sexual reoffense. For example, an offender could have scored high on all actuarial risk assessments and had an extensive history of sexual offending, but if the offender was not found to have a mental abnormality, then that offender could not have been designated for civil commitment. As such, it was only possible for these studies to evaluate the characteristics of offenders designated for civil commitment, not for them to evaluate the importance of just the risk variables to the screening process.

The third analysis of the Florida civil management screening process (Lucken & Bales, 2008), differed from the previous two in that it examined 546 sex offenders screened out at the initial stage of review (i.e., were released) to 246 sex offenders who were referred for the second level of review. As the decision to refer offenders for clinical evaluations was made independent of a finding of mental abnormality, the Lucken and Bales (2008) analysis, therefore, was able to evaluate the importance of demographic, offense, and criminal history variables to the initial screening process. Bivariate comparisons (χ² and t tests) showed significant differences between those offenders who were referred for a face-to-face clinical evaluation and those who were not. Among other findings, referred offenders were older, had more often used a weapon in the commission of their crime, had a greater number of prior sexual and nonsexual offenses, were less likely to have not harmed their victim, and were more likely to select male victims.

Lucken and Bales (2008) then conducted a logistic regression to evaluate the relative influence of the variables on the decision whether to refer or release offenders. The results of this analysis, however, are difficult to interpret given the overall complexity of the model. That is, as regression measures the unique impact of a variable after controlling for (i.e., removing) the variance it shares with all other variables in the model, the regression coefficient for any variable can only be understood within the context of the entire model. For example, offender race was found to be significant in Lucken and Bales’ model (with White offenders being more likely to be referred for a face-to-face clinical evaluation), but only after controlling for the effects of dichotomous offender age (≤30; which was not significant), 25 legal/offense factors (13 of which were significant), 5 victim characteristics (2 of which were significant), and 2 clinical variables (neither of which was significant). The relative impact of any one variable in Lucken and Bales’ regression analysis on the decision of whether or not to refer an offender for clinical evaluation, therefore, can only be interpreted after considering the relative impact of 33 additional factors.

Public safety impact

Although some modern civil management laws have been around for almost two decades, very few empirical evaluations of their impacts have been conducted (Cohen & Jeglic, 2007). This lack of research is understandable given the difficulty in assessing the public safety impact of these laws. That is, by the nature of the laws themselves, almost all of those offenders who get civilly confined are never in the community (only a small percentage of those sex offenders who have been civilly confined have been released; Jackson, Travia, & Schneider, 2008) and, therefore, at risk to sexually reoffend. Thus, it is difficult to determine whether those offenders who become civilly confined are in fact the most likely to sexually recidivate. If the confined offenders are in fact those most likely to sexually recidivate, then it would be reasonable to believe sex offender civil confinement is increasing public safety.

The only three studies on the public safety impact of sex offender civil commitment published to date (Milloy, 2003, 2007; Schram & Milloy, 1998) were conducted by the Washington State Institute for Public Policy. In all three studies, the authors examined the recidivism rates of Washington State sex offenders recommended for civil commitment, but on whose cases the prosecuting attorneys or attorney general decided not to move forward (generally because of difficulty proving one or more of the three criteria for civil commitment required by law). The sexual recidivism rates for the samples in all three studies (range 23%-29%) were considerably higher than the 13%-14% average 5-year rate of sexual rearrest for male sex offenders found in previous research (Hanson & Bussière, 1998; Hanson & Morton-Bourgon, 2004), indicating that the process and criteria used to screen offenders for possible civil commitment in Washington were effectively identifying those offenders most at risk to sexually recidivate. Furthermore, the fact that the samples were comprised of offenders whose cases lacked one or more of the criteria necessary for civil commitment in Washington could mean that those offenders who do become civilly committed (i.e., who meet all three criteria beyond a reasonable doubt) are at an even higher risk to sexually recidivate.

The Current Study

With this in mind, the current analysis attempted to both inform the SOMTA review process and build on the sex offender civil management literature by examining the civil management screening process in NYS. Specifically, the current study first sought to determine which risk-assessment variables most strongly contributed to sex offenders receiving a psychiatric examination. The study would therefore contribute to the literature in two ways: (a) by using a psychiatric exam referral as the dependent measure instead of a recommendation for civil management (thereby isolating the risk markers considered independent of a finding of mental abnormality), and (b) limiting the model to only those variables found to be significant to the process (thereby increasing ease of model interpretation over that of Lucken & Bales, 2008).

Second, the current study sought to determine whether the screening process is, in fact, accurately identifying those offenders most likely to sexually recidivate. As SOMTA had only been enacted a year and a half at the time of this study, however, a sufficient enough length of time had not yet passed for the sexual recidivism rates of those offenders screened under SOMTA to be adequately judged (Hanson & Bussiere, 1998). Thus, the recidivism analyses took a different approach to previous studies of civil commitment impact by using a historical cohort of NYS sex offenders matched on variables found to be important to the SOMTA review process to those offenders actually reviewed under SOMTA (i.e., the offenders used in the recidivism analyses were never reviewed under SOMTA themselves, but were matched on risk factors to SOMTA reviewed offenders).

Method

Two separate samples were required to conduct the analyses. First, a sample of offenders reviewed under SOMTA was needed to examine which variables contributed significantly to the SOMTA review outcome. Second, a historical sample of offenders released before the passage of SOMTA was then needed to be matched to the SOMTA cohort on the variables found to be important to the SOMTA review process to examine recidivism rates.

Samples

SOMTA sample

Demographic and index offense variables for all offenders reviewed for possible civil management (see SOMTA, 2007) are stored in OMH’s Sex Offender Management System (SOMS). As SOMS does not record full criminal histories, however, the computerized criminal history file (which contains information regarding the crimes, charges, and dispositions related to all arrests) for each offender for whom OMH made a final decision regarding possible civil management between April 13, 2007 and November 12, 2008 (N = 1,991) was extracted and provided by the NYS Division of Criminal Justice Services (DCJS). These files were used to calculate offending histories prior to the offender’s SOMTA-qualifying offense.

The SOMTA review process results in each offender either being recommended for civil management or screened out. These civil management recommendations only occur after an offender has been (a) deemed high risk to reoffend sexually, (b) given a psychiatric exam, and (c) found to suffer from a mental abnormality that predisposes the offender to sexually reoffend (SOMTA, 2007). As can be seen in Table 1, 1,711 of the 1,991 offenders reviewed (85.9%) were never deemed high-enough risk to be sent for a psychiatric exam, and only 212 (10.7%) were ultimately recommended for civil management. The remaining 68 offenders (3.4%) were deemed high risk and sent for a psychiatric exam, but were not recommended for civil management.

Table 1.

SOMTA Cohort Characteristics (N = 1,991)

	No psychiatric exam (n = 1,711)		Psychiatric exam (n = 68)		Recommended for civil management (n = 212)
Variables	M (SD)	n (%)	M (SD)	n (%)	M (SD)	n (%)
Age at release^a	38.81 (11.75)		41.13 (11.20)		43.37 (11.20)
Convictions prior to target release
Any^a b	3.91 (3.90)		4.60 (3.28)		6.22 (6.35)
Felony^a	1.90 (1.32)		2.03 (1.13)		2.44 (1.42)
Violent felony^a b	0.93 (0.76)		1.00 (0.88)		1.36 (0.86)
Sexual^a b c	1.09 (0.53)		1.46 (0.80)		2.16 (2.26)
Sexual crime of target release^a b c
Rape		671 (39.22)		29 (42.65)		69 (32.55)
Sexual abuse		479 (28.00)		14 (20.59)		82 (38.68)
Criminal sexual act		311 (18.18)		10 (14.71)		37 (17.45)
Burglary		37 (2.16)		2 (2.94)		9 (4.25)
Robbery		51 (2.98)		2 (2.94)		1 (0.47)
Child sexual contact/performance		78 (4.56)		3 (4.41)		4 (1.89)
Assault		34 (1.99)		0 (0.00)		2 (0.94)
Kidnapping		6 (0.35)		0 (0.00)		3 (1.42)
Homicide		6 (0.35)		3 (4.41)		2 (0.94)
Other		38 (2.22)		5 (2.86)		3 (1.42)
Static-99 score^a c	2.51 (1.62)		5.55 (1.67)		5.92 (1.66)
Five most frequent psychiatric diagnoses^d
Antisocial personality disorder						178 (83.96)
Pedophilia						140 (66.04)
Paraphilia/sexual disorder NOS						114 (53.77)
Alcohol dependence						65 (30.66)
Personality disorder NOS						49 (23.11)

Note: CM = Civil management. Significant between-group differences for continuous variables were assessed through a one-way MANOVA with post hoc two-group comparisons subject to a Bonferroni adjustment (to account for the increased possibility of committing a Type I error), while significant between-group differences for categorical variables were assessed through chi-square and also subject to a Bonferroni adjustment.

Indicates a significant difference between those offenders who did not receive a psychiatric exam and those who were recommended for civil management.

Indicates a significant difference between those offenders who received a psychiatric exam but were not recommended for civil management and those who were recommended for civil management.

Indicates a significant difference between those offenders who did not receive a psychiatric exam and those who received a psychiatric exam but were not recommended for civil management.

As these categories are not mutually exclusive (i.e., it is possible for offenders to have more than one psychiatric diagnosis), presented percentages are of offenders recommended for civil management.

Table 1 displays and compares offender and offense characteristics of the SOMTA cohort across review outcome (condensed characteristics of the entire SOMTA cohort can be found in Table 2). As can be seen, those offenders who were not deemed high-enough risk to receive a psychiatric examination tended to be younger, have less extensive criminal histories, and were scored lower on the Static-99 than offenders who received a psychiatric examination. Likewise, offenders given a psychiatric examination but not recommended for civil management tended to have less extensive criminal histories than those recommended for civil management, but the two groups did not differ in terms of age or Static-99 score. It should be noted, however, that the differences in prior convictions between these two latter groups was largely from two offenders who were recommended for civil management having had very extensive criminal histories.

Table 2.

Characteristics of Offenders across Sample

	SOMTA cohort (N = 1,991)		Potential historical matches (N = 4,807)		Matched historical cohort (n = 1,546)
Variables	M (SD)	n (%)	M (SD)	n (%)	M (SD)	n (%)
Age at release^a b	39.38 (11.80)		38.62 (11.19)		38.24 (10.83)
Convictions prior to target release
Any^a c	4.18 (4.27)		3.68 (3.82)		4.36 (4.07)
Felony^a c	1.96 (1.33)		1.83 (1.23)		1.94 (1.28)
Violent felony	0.97 (0.78)		0.97 (0.74)		0.96 (0.74)
Sexual^b c	1.21 (0.96)		1.18 (0.66)		1.13 (0.49)
Sexual crime of target release^a b c
Rape		769 (38.62)		1,916 (39.86)		655 (42.31)
Sexual abuse		575 (28.88)		1,501 (31.23)		501 (32.36)
Criminal sexual act		358 (17.98)		922 (19.18)		293 (18.93)
Burglary		48 (2.41)		126 (2.62)		19 (1.23)
Robbery		54 (2.71)		105 (2.18)		25 (1.62)
Child sexual contact/performance		85 (4.27)		101 (2.10)		37 (2.39)
Assault		36 (1.81)		69 (1.44)		13 (0.84)
Kidnapping		9 (0.45)		27 (0.56)		2 (0.13)
Homicide		11 (0.55)		24 (0.50)		1 (0.07)
Other		46 (2.31)		16 (0.33)		2 (0.13)
SOMTA screening outcome^b d
No psych exam		1,711 (85.94)				1,377 (88.95)
Psych exam		68 (3.42)				43 (2.78)
Recommended for CM		212 (10.65)				126 (8.14)

Indicates a significant difference between the SOMTA Cohort and the Potential Historical Matches.

Indicates a significant difference between the SOMTA Cohort and the Matched Historical Cohort.

Indicates a significant difference between the Potential Historical Matches and the Matched Historical Cohort.

The SOMTA screening outcomes displayed for the Matched Historical Cohort are not outcomes for the offenders themselves (who were never screened under SOMTA), but for the SOMTA offenders for whom historical matches were found.

Potential historical matches

In all, DOCS identified 4,807 offenders convicted of what would have been SOMTA-qualifying offenses who were released from custody between January 1, 2000 and December 31, 2005. As can be seen in Table 2, the DOCS historical cases tended to be younger and have less extensive criminal histories than offenders in the SOMTA cohort. Therefore, as a whole, it appears the SOMTA cohort was higher risk than the potential historical matches.

Analyses

Backward stepwise logistic regression was used to identify the variables significantly associated with SOMTA review outcome (received a psychiatric exam yes/no), while controlling for other variables in the model. This approach starts with a model including all variables, then excludes variables one-by-one (with the variable with the smallest association to the dependent measure being excluded at each step) until the final model includes only variables significantly associated with the dependent measure. The full model to start the analysis included measures of offender demographics (age at release), risk (Static-99 score), prior criminal history (convictions, felony convictions, drug convictions, violent felony [including violent sexual] convictions, and sexual convictions), and prior criminal justice supervision (prison terms, parole terms, probation terms, supervision violations), as well as victim variables (unrelated, stranger, male).

Instead of using a recommendation for civil management as the dependent measure for the regression, as stated above, a dichotomous measure of whether the offender received a psychiatric exam (0 = no; 1 = yes) was used. This decision was made for several reasons. First, to have received a psychiatric exam, an offender had to be deemed high risk. Second, although those who received a psychiatric exam but were not recommended for civil management differed significantly on several variables from those who were ultimately recommended for civil management, as previously mentioned, those differences shrank or disappeared entirely if two offenders were removed from the analyses.³ Third, mental abnormalities were found during the psychiatric exam for 217 offenders, 212 (97.7%) of which were then recommended for civil management. This high percentage implies the difference between those offenders who received a psychiatric exam but were not recommended for civil management and those who were ultimately recommended was not a function of demographics, criminal history, and/or offense characteristics (i.e., the types of variables used by the SOMTA screeners to judge offender risk, as well as those being tested in the current analysis), but, rather, the mental abnormality. As such, it was felt using whether or not the offender received a psychiatric exam as the outcome measure for the analyses would more accurately assess the SOMTA risk-assessment process.⁴

As discussed in detail below, a matched historical cohort of offenders was then created by matching offenders from the SOMTA cohort to the DOCS historical cases using the variables identified in the backward stepwise logistic regression as being most strongly associated with receiving a psychiatric exam. This allowed for the examination of the 1-, 3-, and 5-year rates of recidivism (any rearrest, felony rearrest, violent felony [including violent sexual] rearrest, sexual rearrest, and sexual reconviction) for the offenders. The matched historical cohort was then broken into groups depending on the SOMTA review outcome of their matched SOMTA case (did not receive a psychiatric exam, received a psychiatric exam) and their rates of sexual rearrest (1-, 3-, and 5-year, as well as survival) were examined.

Results

Process Analysis

Before conducting the backward stepwise logistic regression to determine which demographic and criminal history variables were significantly associated with SOMTA referrals receiving psychiatric examinations, the independent variables in the equation were assessed for levels of multicollinearity. Specifically, possible multicollinearity was assessed by regressing each independent variable on the others, and the regression of prior felony convictions on the other independent variables was the only one to show some signs of multicollinearity (r² ≥ .60). The prior felony convictions variable was, therefore, removed from the model, after which no sign of multicollinearity remained within the analyses as a whole (all r² < .60) or for any variables specifically (all variance inflation factors <2.40).

Results of the backward stepwise regression to identify variables important to the screening process are presented in Table 3. As can be seen, just five variables were retained in the final model, only they were significantly associated with an offender receiving a psychiatric exam after controlling for the influence of the other remaining variables. The four variables which increased the odds of an offender receiving a psychiatric exam were Static-99 score (with each additional Static-99 point increasing the odds of an offender receiving a psychiatric exam by 201%), having a male victim (increased the odds by 140%), prior sexual convictions (each additional prior sexual conviction increased the odds by 109%), and offender age (each year older an offender was at the time of release increased the odds by 4%). Prior violent felony convictions was the only variable negatively associated with an offender receiving a psychiatric exam, with each additional prior violent felony conviction lowering the odds by 28%.⁵

Table 3.

Backward Stepwise Logistic Regression on Referral for a Psychiatric Exam (N = 1,991)

Variables	B	1-Exp(β)^a
Age at release	0.04***	0.04
Static-99 Score	1.10***	2.01
Total prior convictions
Prior violent felony convictions	−0.33*	−0.28
Prior drug convictions
Prior sexual convictions	0.74***	1.09
Prior prison terms
Prior parole terms
Prior probation terms
Prior supervision violations
Unrelated victim
Stranger victim
Male victim	0.88***	1.40
Constant	−8.81***

Note: Coefficients are only presented for variables retained in the final model (p < .05). Variables with dashes were not used in the analysis.

Equal to the change in odds of being referred for a psychiatric exam associated with a one-unit increase in the independent variable.

p < .05. **p < .01. **p < .001.

Recidivism Analysis

Because victim variables and Static-99 scores were not available for the DOCS historical cases (with the lack of victim variables making it impossible to code complete Static-99 scores) and the SOMTA-historical matches were to be made based the regression results, a second backward logistic regression was estimated without the victim variables and with a proxy score for the Static-99 (created by summing offender scores for Static-99 Items 1, 3, 4, 5, 6, and 7) instead of the full Static-99 score.⁶ This model of potential matching variables indicated that only four variables significantly contributed to the likelihood of an offender receiving a psychiatric exam: (a) the Static-99 proxy, (b) prior sexual convictions, (c) offender age at release, and (d) prior drug convictions. As the point of this regression was simply to identify possible variables on which to match SOMTA and historical cases, no odds ratio for this model are discussed and the results are not substantively interpreted.

A cohort of historical cases matched to SOMTA cases was then created using the three variables found in the logistic regression to be most strongly associated with the likelihood of an offender receiving a psychiatric exam (the Static-99 proxy, prior sexual convictions, and offender age at release [within 1 year]), as well as crime of conviction. Also, as rates and patterns of recidivism have been found to vary between male and female sex offenders (Cortoni & Hanson, 2005; Freeman & Sandler, 2008; Sandler & Freeman, 2009) all females screened during the study period (n = 23) were excluded from the matched-cohort analysis. In total, age-, risk-, criminal history-, and crime-based matches could be found for 77.7% of the SOMTA-screened offenders, reducing the final matched historical cohort sample to 1,546.

Characteristics of the matched historical cohort are displayed in Table 2. As can be seen, the SOMTA cohort differed significantly from the matched historical cohort on a number of variables. Specifically, offenders in the SOMTA cohort were found to be older and have more prior sexual convictions than those offenders in the matched cohort. Also, as can be seen in Table 2, historical matches were not found equally for SOMTA cases across the screening outcomes. That is, a significantly higher percentage of historical matches were found for SOMTA offenders who were not referred for a psychiatric exam compared to those who either received a psychiatric exam or were recommended for civil management.

These differences, however, do not indicate poor matching. In fact, comparisons between only those SOMTA offenders for whom matches could be found and the matched historical cohort revealed no significant difference between the two groups on age F(1, 3,091) = 0.15, or prior sexual convictions F(1, 3,091) = 0.00. The observed differences, therefore, indicate that those SOMTA cases for which matches could not be found were higher on certain sexual recidivism risk factors (e.g., prior sexual convictions) and were deemed higher risk according to the risk assessment and record review process.

The 1-, 3-, and 5-year recidivism rates for five different types of recidivism (any rearrest, felony rearrest, violent felony [including violent sexual] rearrest, sexual rearrest, and sexual reconviction) for the matched historical cohort are presented in Table 4. It should be noted, however, that as the SOMTA offenders for whom historical matches could not be found were higher risk than those offenders for whom matches could be found, the recidivism rates presented in Table 4 are probably conservative estimates of what recidivism rates would have been for the entire SOMTA cohort. As can be seen, the 5-year sexual rearrest rate for the entire matched cohort was 7.2%, with those offenders who were rearrested for sexual crimes averaging 3.1 years (SD = 2.1) in the community prior to their first sexual rearrest. The 5-year rates of any rearrest (40.5%), felony rearrest (23.7%), and violent felony (including violent sexual) rearrest (10.5%) were higher than the 5-year rates of sexual rearrest, whereas the 5-year rate of sexual reconviction (5.1%) was lower.

Table 4.

Rates of Recidivism and Time from Release to Recidivism for the Matched Historical Cohort (N = 1,546)

			Recidivism within^a
	Raw total recidivated		1 year (N = 1,546)		3 years (N = 1,544)		5 years (N = 1,184)		Years to recidivism
Recidivism type	n	%	n	%	N	%	n	%	M (SD)	Max
Any rearrest	675	43.6	154	10.0	425	27.5	479	40.5	2.7 (2.0)	8.6
Felony rearrest	423	27.3	065	04.2	223	14.4	280	23.7	3.2 (2.1)	8.6
Violet felony rearrest	178	11.5	033	02.1	090	05.8	124	10.5	3.2 (2.1)	8.6
Sexual rearrest	121	07.8	018	01.2	062	04.0	085	07.2	3.1 (2.1)	8.6
Sexual reconviction	078	05.0	014	00.9	042	02.7	060	05.1	2.8 (1.8)	7.1

The 1-, 3-, and 5-year rates of rearrest were calculated using only those offenders who had at least 1, 3, or 5 years postrelease. Thus, if an offender was first released 1.5 years before the censor date (September 28, 2008) and did not recidivate, he would be counted in the denominator for the 1-year rearrest rate, but not the 3- or 5-year rates.

Table 5 displays the 1-, 3-, and 5-year rates of sexual rearrest for the matched historical cohort broken up by whether or not the SOMTA case to which the historical case was matched received a psychiatric exam. As can be seen, the 1-, 3-, and 5-year rates of sexual rearrest for those historical offenders matched to SOMTA offenders who received a psychiatric exam (4.1%, 7.7%, and 11.6%, respectively) were higher during each time period than those for historical offenders matched to SOMTA offenders who did not receive a psychiatric exam (0.8%, 3.6%, and 6.6%, respectively). Comparisons of the survival curves (see Figure 2) showed these differences to be significant (Wilcoxon-Gehan D = 6.56, p = .01), meaning those historical cases matched to SOMTA cases who received psychiatric exams had significantly higher rates of sexual rearrest than those historical cases matched to SOMTA cases who did not receive a psychiatric exam. The SOMTA review process, therefore, appears to be accurately identifying those offenders at highest risk for sexual rearrest.

Table 5.

Rates of Sexual Re-arrest for the Matched Historical Cohort by SOMTA Review Outcome

			Rearrested within^a						Number of priors
	Raw total rearrested		1 Year		3 Years		5 Years		Any	VFO	Sex
SOMTA review outcome	N	%	n	%	n	%	n	%	M (SD)	M (SD)	M (SD)
No psych exam	100	07.3	11	0.8	49	3.6	69	6.6	4.2 (4.0)	0.9 (0.7)	1.1 (0.4)
Psych exam	021	12.4	7	4.1	13	7.7	16	11.6	6.0 (4.6)	1.3 (0.9)	1.6 (0.8)
Total sample	121	07.8	18	1.2	62	4.0	85	7.2	4.4 (4.1)	1.0 (0.7)	1.1 (0.5)

The 1-, 3-, and 5-year rates of re-arrest were calculated using only those offenders who had at least 1, 3, or 5 years postrelease. Thus, if an offender was first released 1.5 years before the censor date (April 12, 2008) and did not recidivate, he would be counted in the denominator for the 1-year rearrest rate, but not the 3- or 5-year rates. Thus, the base number of offenders in the 1-, 3-, and 5-year rearrest rates varies.

Discussion

The current study sought to investigate the functioning and efficacy of the sex offender risk assessment and record review process in New York State. Specifically, in the first part of the analysis, the study investigated which demographic, criminal history, and victim variables contributed most significantly to an offender being deemed high risk and receiving a psychiatric exam. The study then used the results of this analysis to create a sample of historical cases matched to cases actually reviewed under SOMTA on the variables found to be significant in the risk-assessment process. Finally, the study investigated the recidivism patterns of the matched historical cases to evaluate whether or not the risk assessment and record review process in NYS was effective in identifying which offenders were most likely to recidivate.

Although between-group tests revealed significant differences on almost every variable of interest between those SOMTA reviewed offenders who did not receive a psychiatric exam (n = 1,711), those who received a psychiatric exam but were not recommended for civil management (n = 68), and those who were ultimately recommended for civil management (n = 212), when the influence of other variables was controlled for through regression, only five variables were found to significantly differentiate those who received a psychiatric exam from those who did not. Four of these five variables (Static-99 score, having a male victim, prior sexual convictions, and offender age at release) were found to increase the odds of an offender receiving a psychiatric exam, whereas one variable (prior violent felony [including violent sexual] convictions) lowered the odds. The greatest increase in the odds of an offender receiving a psychiatric exam was associated with the offender’s Static-99 score, with each additional point on the Static-99 raising the odds of a psychiatric exam by 201%.

The fact that Static-99 score was positively associated with the odds of an offender receiving a psychiatric exam was not surprising given the key role it plays in the early stages of the risk assessment and record review process (see Figure 1). Even after controlling for the influence of the Static-99, however, four other variables (having a male victim, prior sexual convictions, offender age at release, and prior violent felony convictions) all of which are included on the Static-99 still significantly influenced the SOMTA screening process. Interestingly, two of these variables (having a male victim and prior sexual convictions) were associated with the SOMTA screening outcome in the same direction as they contribute to the Static-99 total score (and also risk of sexual recidivism; Hanson & Bussière, 1998; Hanson & Morton-Bourgon, 2004), whereas the other two variables (offender age at release and prior violent felony convictions) had opposite associations. These findings may imply that, after taking into account the Static-99 total score, certain factors included on the Static-99 are then separately given more and less consideration by OMH staff conducting the reviews. An alternative explanation for the inverted impact of offender age at release and prior violent felonies on the screening outcome, however, may be that the risk-assessment process targets child molesters (as evidenced by 66% of all those recommended for civil management having been diagnosed as pedophiles [see Table 1]), who tend to be older on average than other sex offenders (Hanson, 2002), offend later into their lifespan (Hanson, 2002), and offend less violently (Lanyon, 1986).

The recidivism analysis using historical offenders matched to SOMTA reviewed offenders on age (within 1 year), Static-99 proxy score, number of prior sexual convictions, and crime of conviction revealed an overall 5-year sexual rearrest rate of 7.2% (see Table 4). When the analysis separated the historical offenders whose SOMTA match received a psychiatric exam from those whose match did not (see Table 5), the historical offenders whose SOMTA match received a psychiatric exam recidivated at a significantly higher rate. This difference in recidivism rate was consistent throughout the 5-year period, with the psychiatric exam matches having a 5-year sexual rearrest rate 75.8% higher than the no psychiatric exam matches. Although the raw difference in 5-year sexual recidivism rate (5.0%) was not as large as expected, this may be because of to the data limitations discussed below.

Taken as a whole, the results of this study support the efficacy of the risk assessment and record review process used by OMH staff to identify those offenders most likely to recidivate sexually. Specifically, the results indicate that the factors most likely to be associated with an offender being deemed high risk and receiving a psychiatric exam are those empirically derived in the research literature as being significantly related to sexual recidivism (e.g., Hanson & Morton-Bourgon, 2004), and that offenders who share risk factors with those who receive psychiatric exams recidivate at a significantly higher rate.

Limitations

One limitation of the current study was that of missing data. Specifically, the lack of victim information for the historical cohort made it impossible to match the offenders to SOMTA cases on all variables significantly associated with an offender receiving a psychiatric exam (e.g., male victim). The current study, however, accounted for this fact by running a second analysis to identify which available variables most strongly predicted an offender receiving a psychiatric exam and then matching based on those results. The lack of psychiatric diagnoses, furthermore, made it impossible to match historical offenders to SOMTA offenders who were ultimately recommended for civil management on particular mental abnormalities. This limitation means the current study could only examine the ability of the risk-assessment process to identify high risk offenders before the psychiatric exam, not if those offenders who receive civil management (i.e., are found to have a mental abnormality) are more likely to sexually recidivate than those who do not. These two questions, however, are closely related (as evidenced by 74.6% of those offenders who received a psychiatric exam having eventually been recommended for civil management), and the results of the present study strongly indicate that offenders recommended for civil management probably recidivate sexually at a higher rate than those not recommended.

Also related to the matching, the current study is limited in that historical matches could not be found for some of the highest risk SOMTA offenders. This is shown in Table 2, where the SOMTA cohort had a significantly larger percentage of offenders referred for a psychiatric exam (including those eventually referred for civil management) than the matched-historical cohort. This fact means the 5-year sexual rearrest rate observed in the matched cohort is possibly an underestimate of what would have been the 5-year sexual recidivism rate for the SOMTA cohort as a whole (had the entire SOMTA cohort been released), as well as for those SOMTA offenders referred for a psychiatric exam (had they all been released). This is one possible reason for the significantly higher sexual recidivism rate found for released offenders in the Washington State studies (23%-29%; Milloy, 2003, 2007; Schram & Milloy, 1998) than for those offenders matched to offenders who received a psychiatric exam in the present study (roughly 12% over 5 years).

Another related limitation of the current study was that examining the recidivism of the SOMTA reviewed offenders themselves was impossible. This limitation was the result of two factors: (a) offenders deemed to be the highest risk having been civilly confined (i.e., they were not at risk in the community during the study period), and (b) little follow-up time (less than 2 years) for those offenders who were released (i.e., were screened out by the process). These two facts mandated the use of a historical sample matched to the SOMTA reviewed cohort. Although the present study took steps to make sure the offenders were matched on variables important to the review outcome, it is likely that the samples were not matched on all important risk factors (e.g., emotional identification with children; Hanson & Morton-Bourgon, 2004).

Conclusions

Unlike the growing body of research that indicates little, no, or negative impacts on public safety resulting from sex offender registries and community notification (Sandler, Freeman, & Socia, 2008; Zgoba, Witt, Dalessandro, & Veysey, 2008), as well as sex offender residency restrictions (Colorado Department of Public Safety, 2004; Duwe, Donnay, & Tewksbury, 2008), results of the current study (and the studies from Washington) indicate that sex offender civil management is likely increasing public safety. This finding is not terribly surprising, seeing as the screening procedures used in NYS to select those offenders recommended for civil management were designed around the empirical risk-assessment research. Thus, unlike sex offender registries, community notification, and residency restrictions, sex offender civil management appears to likely be reducing sexual victimizations.

The question remains, however, as to whether these public safety gains are worth the tremendous costs associated with sex offender civil management, specifically inpatient civil confinement. According to the most recent annual report from OMH, inpatient civil confinement costs NYS approximately US$175,000 per offender per year (New York State Office of Mental Health, 2009). To civilly confine 100 offenders for 5 years, therefore, it would cost roughly US$87.5 million. Although the confinement ensures these offenders will not commit another offense in the community over that period (for context, the average 5-year sexual recidivism rate for high-risk sex offenders is believed to be around 30%), it remains to be seen whether this time in confinement and its associated time in treatment will significantly impact sexual recidivism rates for these offenders following their release. Likewise, the public safety impact of outpatient civil management (SIST) has not yet been adequately evaluated. Research needs to be conducted, therefore, to examine the sexual recidivism rates of offenders placed directly onto SIST (as well as those released from confinement). If the sexual recidivism rate for offenders placed directly onto SIST is found to be low, SIST may present a much less expensive, yet still effective, management alternative to confinement.

The fact remains, however, that all public policies targeting previously convicted sex offenders are limited in their ability to substantially reduce instances of sexual offending. That is, research has consistently found that the vast majority of sexual offenses are committed not by previously convicted sex offenders, but by offenders with no prior criminal history of sexual offending (Greenfeld, 1997; Sandler et al., 2008). For example, Sandler et al. (2008) recently found that over a 21-year span, less than 5% of all arrests for registerable sexual offenses in NYS were of offenders previously convicted of a sexual offense. From a purely numeric standpoint, this finding means efforts directed at the prevention of first time sexual offenses have more potential to substantially reduce sexual offending than policies directed at previously convicted sex offenders. It may be, therefore, that funds allocated to inpatient civil confinement could have a greater impact on reducing sexual offending if they were directed toward such primary prevention programs as community awareness initiatives (much like those offered by Stop It Now!).

Footnotes

The opinions contained within the article are those of the researchers alone, and they do not reflect the opinions of the New York State Office of Mental Health.

The author(s) declared no potential conflicts of interest with respect to the authorship and/or publication of this article.

The author(s) received no financial support for the research and/or authorship of this article.

Notes

References

Archer

R. P.

Buffington-Vollum

J. K.

Stredny

R. V.

Handel

R. W.

(2006). A survey of psychological test use patterns among forensic psychologists. Journal of Personality Assessment, 87, 84-94.

Cohen

Jeglic

E. L.

(2007). Sex offender legislation in the United States: What do we know? International Journal of Offender Therapy and Comparative Criminology, 51, 369-383.

Colorado Department of Public Safety. (2004). Report on safety issues raised by living arrangements for and location of sex offenders in the community. Denver, CO: Author.

Cortoni

Hanson

R. K.

(2005). A review of the recidivism rates of adult female sexual offenders. Ottawa, Ontario: Correctional Service Canada.

Duwe

Donnay

Tewksbury

(2008). Does residential proximity matter: A geographic analysis of sex offense recidivism. Criminal Justice and Behavior, 35, 484-504.

Freeman

N. J.

Sandler

J. C.

(2008). Female and male sex offenders: A comparison of recidivism patterns and risk factors. Journal of Interpersonal Violence, 23, 1394-1314.

Greenfeld

(1997). Sex offenses and offenders: An analysis of data on rape and sexual assault. Washington, DC: Bureau of Justice Statistics.

Hanson

R. K.

(2002). Recidivism and age: Follow-up data from 4,673 sexual offenders. Journal of Interpersonal Violence, 17, 1046-1062.

Hanson

R. K.

Bussière

M. T.

(1998). Predicting relapse: A meta-analysis of sexual offender recidivism studies. Journal of Consulting and Clinical Psychology, 66, 348-362.

10.

Hanson

R. K.

Morton-Bourgon

(2004). Predictors of sexual recidivism: An updated meta-analysis (Report 2004-02). Gatineau, Quebec: Public Works and Government Services Canada.

11.

Hanson

R. K.

Morton-Bourgon

(2009). The accuracy of recidivism risk assessments for sexual offenders: A meta-analysis. Psychological Assessment, 21, 1-21.

12.

Hanson

R. K.

Thornton

(2000). Improving risk assessments for sex offenders: A comparison of three actuarial scales. Law and Human Behavior, 24, 119-136.

13.

Harris

Phenix

Hanson

R. K.

Thornton

(2003). STATIC-99 coding rules revised 2003. Ottawa, Ontario: Department of the Solicitor General of Canada.

14.

Jackson

R. L.

Hess

D. T.

(2007). Evaluation for civil commitment of sex offenders: A survey of experts. Sex Abuse, 19, 425-448.

15.

Jackson

Travia

Schneider

(2008). Annual survey of sex offender civil commitment programs. Sex Offender Civil Commitment Programs Network (SOCCPN). Retrieved from http://soccpn.org/images/2008_20Annual_20Survey_20of_20Programs.pdf

16.

Lanyon

R. I.

(1986). Theory and treatment in child molestation. Journal of Consulting and Clinical Psychology, 54, 176-182.

17.

Levenson

J. S.

(2004). Sexual predator civil commitment: A comparison of selected and released offenders. International Journal of Offender Therapy and Comparative Criminology, 48, 638-648.

18.

Levenson

J. S.

Morin

J. W.

(2006). Factors predicting selection of sexually violent predators for civil commitment. International Journal of Offender Therapy and Comparative Criminology, 50, 609-629.

19.

Lucken

Bales

(2008). Florida’s sexually violent predator program: An examination of risk and civil commitment eligibility. Crime & Delinquency, 54, 95-127.

20.

McGrath

R. J.

Cumming

G. F.

Burchard

B. L.

(2003). Current practices and trends in sexual abuser management: The Safer Society 2002 Nationwide Survey. Brandon, VT: Safer Society Foundation.

21.

Milloy

(2003). Six-Year Follow-Up of Released Sex Offenders Recommended for Commitment Under Washington’s Sexually Violent Predator Law, Where No Petition Was Filed. Olympia,WA: Washington State Institute for Public Policy.

22.

Milloy

(2007). Six-Year Follow-Up of 135 Released Sex Offenders Recommended for Commitment Under Washington’s Sexually Violent Predator Law, Where No Petition Was Filed. Olympia,WA: Washington State Institute for Public Policy.

23.

New York State Office of Mental Health. (2009). 2008 Annual Report on the Implementation of Mental Hygiene Law Article 10: Sex Offender Management and Treatment Act of 2007. Albany, NY: Author.

24.

Sandler

J. C.

Freeman

N. J.

(2009). Female sex offender recidivism: A large-scale empirical analysis. Unpublished Manuscript.

25.

Sandler

J. C.

Freeman

N. J.

Socia

K. M.

(2008). Does a watched pot boil? A time-series analysis of New York State’s Sex Offender Registration and Notification, Psychology, Public Policy, and Law, 14, 284-302.

26.

Schram

Milloy

C. D.

(1998). Sexually violent predators and civil commitment: A study of the characteristics and recidivism of sex offenders considered for civil commitment but for whom proceedings were declined. Olympia: Washington State Institute for Public Policy.

27.

Sex Offender Management and Treatment Act [SOMTA], Mental Hygiene Law, §2, Article 10 (2007).

28.

Zgoba

Witt

Dalessandro

Veysey

(2008). Meghan’s Law: Assessing the practical and monetary efficacy. Washington, DC: U.S. Department of Justice.