Gants, J.
The Commonwealth of Massachusetts has petitioned this Court to commit the respondent, Wilson Rodriguez, as a sexually dangerous person under G.L.c. 123A. A probable cause hearing was conducted on March 15 and 20, 2001 as required under G.L.c. 123A, §12(c), at which the Commonwealth called one expert witness, Dr. William B. Land, and Rodriguez called another expert witness, Dr. Daniel Kriegman. The Commonwealth offered into evidence three exhibits that included the police reports and Department of Social Services file regarding Rodriguez’s prior convictions for rape and indecent assault of a child, his Board of Probation record, and his Department of Correction (“DOC”) file. Rodriguez offered into evidence seven exhibits, including three learned treatise articles on the subject of predicting recividism by sexual offenders. Having considered the testimony at the probable cause hearing and the exhibits offered into evidence at that hearing, this Court finds that there is not probable cause to believe that Rodriguez is a sexually dangerous person as defined in G.L.c. 123A, §1. Since Rodriguez has completed his term of imprisonment and is being held in custody only as a result of this petition, this Court ORDERS that Rodriguez be released from custody forthwith and commence his term of probation.
“Probable Cause” Under the “Directed Verdict” Standard
In Commonwealth v. Lawrence A. Bruno, the Supreme Judicial Court declared that the “sufficient showing” that permits a Court temporarily to detain a respondent who is scheduled to be released from prison is probable cause to believe that the respondent is a sexually dangerous person within the meaning of meaning of G.L.c. 123A, §1. 432 Mass. 489, 510-11 (2000). The Supreme Judicial Court in Bruno also declared that the standard for a “probable cause” hearing under G.L.c. 123A, §12(c) is not the probable [28]*28cause standard but rather the “directed verdict” standard used in probable cause bind-over hearings under G.L.c. 276, §38. Id. at 510. Before this Court can consider whether the evidence at the probable cause hearing supports a finding of “probable cause” under the “directed verdict” standard set forth in Bruno, this Court must first determine what the “directed verdict” standard is and how it compares to the traditional probable cause standard.
In Bruno, the Supreme Judicial Court both quoted and cited with approval the case of Myers v. Commonwealth, 363 Mass. 843 (1973), which interpreted the meaning of probable cause in the context of a District Court hearing to bind over to the Superior Court a defendant accused of serious felonies. Bruno, 432 Mass. at 509-10. From Myers, one can discern that the determination under G.L.c. 123A, § 12(c) of whether probable cause exists to believe that the defendant is a sexually dangerous person:
1. requires more evidence than probable cause to arrest but less than proof beyond a reasonable doubt; Myers at 850; Bruno at 510;
2. requires the factfinder to evaluate the credibility of witnesses and the quality of the evidence introduced; Myers at 853; and
3. is analogous to the court’s ruling on a motion for a directed verdict in that ”[t]he examining magistrate should view the case as if it were a trial and he were required to rule on whether there is enough credible evidence to send the case to the jury.” Myers at 850.
It is plain from Myers that the “directed verdict” standard is simply “analogous” to the civil directed verdict standard and not identical to it. Id.. If the standards were identical, it would require the factfinder at a probable cause hearing under G.L.c. 123A, § 12(c) to view the evidence in the light most favorable to the prosecution and determine whether the evidence viewed in that light is sufficient reasonably to support a finding that the defendant is a sexually dangerous person. See Cambridgeport Savings Bank v. Boersner, 413 Mass. 432, 438 (1992). Under the traditional directed verdict standard, the court is not permitted to weigh the credibility of the witnesses or otherwise consider the weight of the evidence. Id. If this civil “directed verdict” standard were to be used in probable cause determinations under G.L.c. 123A, § 12(c), probable cause could be found in cases that would not pass muster under the traditional probable cause standard because the court is permitted to consider credibility when it determines traditional probable cause and would be barred from this consideration in determining “directed verdict” probable cause. This would mean that a court, on identical evidence, could find that there was not traditional probable cause to justify temporary detention but that there was “directed verdict” probable cause to hold the defendant in the treatment center pending trial. Indeed, it would mean that a court must find probable cause to believe the defendant to be sexually dangerous whenever the Commonwealth presents evidence, regardless of how incredible, that would be sufficient, if believed, to find beyond a reasonable doubt that the defendant is a sexually dangerous person. In Myers, the Supreme Judicial Court made it clear that this would not be enough to bind over a defendant to Superior Court: there must be “credible evidence” sufficient to permit a rational trier of fact to find the defendant guilty. Myers at 850.
Therefore, this Court concludes that the “directed verdict” standard of probable cause set forth in Bruno requires this Court first to determine the evidence it finds credible and then, considering only that credible evidence, determine whether a reasonable trier of fact could find the defendant beyond a reasonable doubt to be a sexually dangerous person based on that credible evidence.
Findings of Fact
Rodriguez is a 65-year-old male who has been in custody since August 1996. He was married, had children, became divorced, and then lived with a woman who was the mother of a young girl (who I shall refer to as “S," born on October 26, 1983) and two boys. Rodriguez moved in with this woman and, according to S in her SAIN interview, raised her since she was seven or eight years of age. S told the investigators that she considered him her stepfather.
In December 1995, S’s brother came home from school and saw Rodriguez, wearing only boxer shorts, in bed on top of S, who was 12 years old. The brother called the police. S declared in the SAIN interview that all they were doing in bed was tickling each other but forensic examination disclosed that there was seminal fluid containing sperm on S’s underwear. Rodriguez was charged in District Court with indecent assault and battery on a child under 14, and was released on bail, with a special condition to stay away from S.
On July 16, 1996, while the indecent assault and battery charge remained pending, Rodriguez was observed with S at a convenience store in East Bridgewater. Witnesses observed unusual sexual contact between this 60-year-old man and 12-year-old girl — they were kissing, holding hands, and S even grabbed Rodriguez’s genitals as they were leaving the store. S had come to the store by bicycle and Rodriguez had driven there in a van. Rodriguez put the bicycle in his van, and S followed the bicycle inside the van. They drove to the rear of a nearby business, followed covertly by the convenience store manager. When the manager no longer saw their heads in the van, he telephoned the police, who soon arrived. Rodriguez and S tried to run in different directions when the police announced themselves, but they were soon apprehended.
When interviewed, S admitted that she and Rodriguez had been having vaginal intercourse at least a couple of times each week in his van for the last couple of months.
[29]*29S also said that Rodriguez had touched her on her breasts and that she had touched him on his genitals. S said that her mother had been sent to jail, and that the sexual relationship began about two weeks after that. (Using the date that S’s mother began her jail term as a parameter, the sexual intercourse continued for roughly three months.) S described their relationship as a boyfriend-girlfriend relationship, which began when he asked her if she wanted to have sex and she said yes.
Rodriguez has been held in custody since his arrest on July 16, 1996. With respect to the three months of sexual misconduct between May and July 1996, Rodriguez was indicted on one count of rape of a child with force, in violation of G.L.c. 265, §22A, and two counts of indecent assault and battery of a child under 14 years old.1 He was also indicted on one count of indecent assault and battery of a child under 14 with respect to the December 1995 incident. On June 25, 1998, after 707 days in custody, Rodriguez pleaded guilty to all the pending indictments and was sentenced to serve 4-5 years at MCI Cedar Junction, with five years probation to commence on and after his release from prison. As conditions of probation, Rodriguez must participate in sex offender counseling as directed by the Probation Department and must have no contact with S.2
While Rodriguez has a modest criminal record apart from these convictions, there are no other charges that allege sexual misconduct. In prison, his record was nearly unblemished; he received two disciplinary reports, one for failing to keep his quarters in order and another for possessing items belonging to another prisoner. He refused, however, to pursue the Sex Offender Treatment Program (“SOTP”).
There is a notation in his DOC file that he “has alcohol related abuse issues,” but there is no indication of what this means or where this information came from. His Board of Probation record reveals that in 1990 he was fined for drinking alcohol in public. In 1989, he was charged with operating under the influence of alcohol but he was not convicted of this charge. There is nothing in either the DSS file or the police reports to indicate that he was abusing alcohol in 1995 and 1996. Indeed, in the DSS file, S’s brother denied seeing alcohol in the home or knowing anyone who used alcohol.
Conclusions of Law
To be found a sexually dangerous person as defined in G.L.c. 123A, §1, the Commonwealth must meet its burden of showing that Rodriguez (1) has been convicted of a sexual offense, as defined in G.L.c. 123A, §1; (2) suffers from a “mental abnormality or personality disorder”; and (3) that this mental abnormality or personality disorder makes him “likely to engage in sexual offenses if not confined to a secure facility.”3 There is no dispute that Rodriguez, having been convicted of three counts of indecent assault and battery on a child under 14 and one count of rape by force of a child under 16, has been convicted of a sexual offense. With respect to the second and third elements, however, the Commonwealth’s case rests solely on the testimony of its expert, Dr. Land.
Dr. Land is Board certified in psychiatry, with subspecialty certifications in forensic psychiatry, geriatric psychiatry, and addiction psychiatry. From 1992 through 1999, he was a supervisor in the Department of Psychiatry at Beth Israel Hospital, a consulting psychiatrist with the Massachusetts Rehabilitation Commission, and a staff psychiatrist at Bridgewater State Hospital. During this time period, he also worked for roughly two years at the Massachusetts Treatment Center at Bridgewater for Sexually Dangerous Offenders, and treated roughly 200 sexual offenders.
1. Does Rodriguez Suffer From A Personality Disorder?
Dr. Land opined that Rodriguez suffered from a single personality disorder — pedophilia, as defined in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision, published in 2000 by the American Psychiatric Association, commonly referred to as DSM-IV. (In this decision, this Court shall cite it as DSM-IV-TR, to make clear that the Court is referring to the Text Revision published in 2000 and not the earlier DSM-IV published in 1994.) Dr. Land stated that Rodriguez was not mentally ill and did not suffer from any mental abnormality. His finding that Rodriguez was a pedophile was based solely on Rodriguez’s sexual offenses in 1995 and 1996; he did not interview him or examine any psychiatric or psychological records (if any exist).
Under G.L.c. 123A, §1, “personality disorder” is defined as “a congenital or acquired physical or mental condition that results in a general lack of power to control sexual impulses.” Simply because a person suffers from a “mental disorder” as defined in DSM-IV does not automatically mean that the person suffers from a “personality disorder” as defined in G.L.c. 123A, §1. Indeed, the Introduction to DSM-IV-TR specifically warns:
When the DSM-IV categories, criteria, and textual descriptions are employed for forensic purposes, there are significant risks that diagnostic information will be misused or misunderstood. These dangers arise because of the imperfect fit between the questions of ultimate concern to the law and the information contained in a clinical diagnosis. In most situations, the clinical diagnosis of a DSM-IV mental disorder is not sufficient to establish the existence for legal purposes of a “mental disorder,” “mental disability,” “mental disease,” or “mental defect.”
DSM-IV-TR, Introduction at xxii-xxiii. Since the DSM-IV warns that a person suffering from a “mental disorder” under the DSM-IV may not suffer from a “mental disorder” as defined in law, one would presume that the warning would be even more forceful if the legal classification were a “personality disorder” whose definition [30]*30specifically requires that the mental condition result "in a general lack of power to control sexual impulses.”
The DSM-IV characterizes pedophilia as a type of Paraphilias, which is a general category of sexual disorder. “The essential features of a Paraphilia are recurrent, intense sexually arousing fantasies, sexual urges, or behaviors generally involving 1) nonhuman objects, 2) the suffering or humiliation of oneself or one’s partner, or 3) children or other non-consenting persons, that occur over a period of at least six months.” DSM-IV-TR at 566. For all Paraphilias other than sexual sadism, “the diagnosis is made if the behavior, sexual urges, or fantasies cause clinically significant distress or impairment in social, occupational, or other important areas of functioning.” Id.
The “diagnostic criteria” for pedophilia, according to DSM-IV, are:
2. Over a period of at least 6 months, recurrent, intense sexually arousing fantasies, sexual urges, or behaviors involving sexual activity with a prepubescent child or children (generally age 13 years of younger).
3. The person has acted on these sexual urges, or the sexual urges or fantasies cause marked distress or interpersonal difficulty.
4. The person is at least age 16 years and at least 5 years older than the child or children in Criterion A.
DSM-IV-TR at 572. As should be clear from these “diagnostic criteria,” a person may properly be diagnosed as a pedophile under DSM-IV even if that person has never sexually touched a child; it is sufficient that the person have intense and recurrent sexual fantasies or urges towards a prepubescent child that cause either significant distress or interpersonal difficulty. Even if a person all his life successfully represses those fantasies or urges and never acts upon them, that person may still be deemed a pedophile if all three criteria are met. In short, DSM-IV characterizes a person as a pedophile if he is intensely sexually attracted to young children and his sexual urges or fantasies cause him marked distress or interpersonal difficulty, regardless of whether or not he acts on those fantasies or urges.4
There is nothing in DSM-IV, either in the “diagnostic criteria” or in the description of this sexual disorder, that discusses whether or not an intense sexual attraction to children “results in a general lack of power to control sexual impulses,” which is what a mental disorder must result in for it to be a “personalily disorder” under G.L.c. 123A, §1. Rather, the DSM-IV focuses on the intense sexual attraction to children as the sexual disorder, and offers no guidance as to whether or not a pedophile is likely to have the power to control these sexual impulses and not act upon them.
Nor can one say, in the absence of clinical or empirical data, that all pedophiles, by the nature and intensity of their sexual attraction, have a general lack of power to control these sexual impulses. It may indeed be true that all (or most) sex offenders against children are pedophiles, but that does not necessarily lead to the conclusion that all pedophiles are sex offenders. This Court does not know how many pedophiles never act on their sexual impulses towards children, and there was no information proffered by Dr. Land to shed light on this question. The DSM-IV reflects that some percentage of pedophiles do not act on their urges because it declares, “Individuals with Pedophilia who act on their urges with children may limit their activity to undressing the child and looking, exposing themselves, masturbating in the presence of the child, or gentle touching and fondling of the child.” DSM-IV-TR at 571 (emphasis added). If all pedophiles acted on their urges, then the italicized language would have been unnecessary. Nor can one simply assume from common sense that pedophiles have a general lack of power to control their sexual impulses towards children; many heterosexuals have intense and recurrent sexual fantasies and urges towards adults of the opposite sex, but one would not assume among this population a general lack of power to control these sexual impulses.
Nor could Dr. Land reasonably declare from his clinical experience that pedophilia “results in a general lack of power to control sexual impulses.” Dr. Land conceded that he had been involved in treating roughly 200 sexual offenders during his two years working at the Bridgewater Treatment Center, some share of which were pedophiles, but he conducted no follow-up of these patients and did not know what happened to them after their release. To his credit, Dr. Land recognized that his interviews with these patients, without any follow-up, does not reasonably permit him from his own clinical experience to offer any informed opinion as to the likelihood that pedophiles as a group would again commit sexual offenses. Since the pedophiles at the Treatment Center were among the subset of pedophiles who had not controlled their sexual impulses, this means that Dr. Land, focusing only on this subset of pedophiles, could not from his own clinical experience even offer any opinion as to their rate of recidivism for sexual crimes.
Dr. Land, however, did contend that empirical studies have demonstrated a high rate of recidivism for such offenders, although he could not identify any study by name or discuss the details of any particular study. He did testify, as did Dr. Kriegman, that R. Karl Hanson was the leading scholar in the field regarding empirical studies of sexual recidivism. Fortunately, three articles authored or co-authored by Hanson were offered into evidence as learned treatises;
1. Hanson, R. Karl, “What Do We Know About Sex Offender Risk Assessment,” Psychology, Public Policy and Law, 1998, Vol. 4, No. , pp. 50-72;
2. R.K. Hanson & Bussiere, M.T., Predictors of Sexual Offender Recidivism: A Meta-Analysis (User Report No. 96-04) Ottawa, Ontario, Canada; Department of the Solicitor General of Canada (1996); and
[31]*313. R.K. Hanson & Thornton, David, Static 99: Improving Actuarial Risk Assessments for Sex Offenders (User Report No. 99-02) Ottawa, Ontario, Canada: Department of the Solicitor General of Canada (1999);
These articles examine and summarize the findings from empirical studies in the research literature regarding the rate of recidivism of sexual offenders. While this Court will discuss these findings in far greater detail later in this Memorandum, it suffices here to observe that, based on a review of 61 studies providing information on a total of 29,972 sexual offenders, of which 9,603 were child molesters, the sex offense recidivism rate over the average 4-5 year follow-up period of these studies for child molesters was 12.7 percent. R.K. Hanson & Bussiere, M.T., Predictors of Sexual Offender Recidivism: A Meta-Analysis at 13. “These averages should be considered cautiously, since they are based on diverse methods and follow-up periods, and, as previously mentioned, many sexual offenses remain undetected. These global figures, nonetheless, provide the general context within which to interpret the effects of the various predictor variables.” Id. While the largest single predictor of sexual offense recidivism in these studies was “a sexual preference for children as measured by phallometric methods,” the correlation coefficient (known in statistics as “r”) for this predictor was .32.5 Id. at 14. This correlation, while plainly substantial, must be viewed in light of the 12.7 percent base rate of recidivism.6 In short, while a previous act of child molestation raises a significant risk that the sexual offender will commit new sexual offenses, especially when phallometric methods measure a sexual preference for children, the recidivism rate is not so high that one can fairly conclude that every pedophile (or even most pedophiles) who has been convicted of a sexual offense has “a general lack of power to control sexual impulses” resulting from his pedophilia. The empirical evidence demonstrates that, while the sexual attraction to children will not likely diminish in a pedophile after release from custody, many (and probably a majority) of pedophiles do not criminally act upon those impulses within a 4-5 year time frame.
In light of this analysis, it is plain that Dr. Land has done precisely what the DSM-IV warned against; he has mistakenly and misleadingly equated a mental/sexual disorder under the DSM-IV with the legal definition of “personality disorder” when it does not properly fit. To meet the definition of “personality disorder” under G.L.c. 123A, §1, it is not enough simply to find that the respondent is a pedophile as defined in the DSM-IV; one must also find that the pedophilia “results in a general lack of power to control sexual impulses,” which requires the Court to go beyond the diagnostic criteria for pedophilia under the DSM-IV. Indeed, to meet this legal definition, one must go beyond the DSM-IV entirely, because this Court is aware of no mental/sexual disorder identified in the DSM-IV that has “a general lack of power to control sexual impulses” as a specified diagnostic criterion. Pragmatically, this means that, to find that a respondent has a “personality disorder” under G.L.c. 123A, §1, one must find that the respondent has a mental/sexual disorder as defined in the DSM-IV and, based on evidence specific to the respondent, that this mental/sexual disorder, for this respondent, "results in a general lack of power to control sexual impulses.”
The importance of this finding, especially for pedophiles, cannot be overestimated. Since a pedophile, under Criterion A of the DSM-IV has “recurrent, intense sexually arousing fantasies, sexual urges, or behaviors involving sexual activity” with prepubescent children, the finding that a pedophile has a general lack of power to control his sexual impulses leads naturally to the conclusion that the pedophile is likely to act on these recurrent, intense sexual impulses by committing a sexual offense against a child unless confined.
Looking at the evidence specific to Mr. Rodriguez, assuming for the moment that he is a pedophile as Dr. Land has opined, there is no credible evidence that there is anything about him apart from his previous sexual offenses with S, that indicates that his pedophilia results in a general lack of power to control sexual impulses. Dr. Land focuses on what he contends to be evidence of Rodriguez’s alcohol abuse, but there is too little evidence in the record to permit any reasonable conclusion that Rodriguez had been abusing alcohol in the years prior to his incarceration. Neither the notation in his DOC file that he “has alcohol related abuse issues,” without any indication of what this means or where this information came from, nor a prior conviction in 1990 for drinking alcohol in public is sufficient to warrant this conclusion, especially when there is nothing in either the DSS file or the police reports to indicate that he was abusing alcohol in 1995 and 1996. In any event, even if Rodriguez had an alcohol problem, Hanson’s survey of the empirical studies found that sexual offense recidivism was unrelated to substance abuse. R.K. Hanson & Bussiere, M.T., Predictors of Sexual Offender Recidivism: A Meta-Analysis at 19.7
Dr. Land also focuses on the fact that Rodriguez refused sex offender treatment, but there is no empirical evidence identified by Dr. Land to support the inference that such refusal is correlated with a greater risk of recidivism. Indeed, Hansen’s review of the 61 studies found that, “(c)ontrary to what is commonly assumed, those sexual offenders who denied their offenses were no higher risk than other offenders (average r of .02, with no significant variability).” R.K. Hanson & Bussiere, M.T., Predictors of Sexual Offender Recidivism: A Meta-Analysis at 14. It is true that, among those offenders who began treatment, those “who were unmotivated to attend treatment, or who failed to complete treatment, were at greater risk for general recidivism than those who completed treatment (r=. 14).” Id. at 16. But there is no evidence in the [32]*32probable cause record to support the opinion that the refusal of treatment is correlated with sexual recidivism.8
There is one other formidable problem with Dr. Land’s opinion that Rodriguez, as a pedophile, suffers from a mental condition that results in a general lack of power to control sexual impulses. The DSM-IV reports in the section regarding pedophilia, “The disorder usually begins in adolescence, although some individuals with Pedophilia report that they did not become aroused by children until middle age.” DSM-IV-TR at 571. Rodriguez was 60 years old before there was any evidence of sexual misconduct with a young girl. This means either that he was a pedophile for many years but was successfully able to control his sexual impulses, or that he was the extraordinary instance of a man who became a pedophile at the age of 60. Since the DSM-IV does not even recognize the possibility of an elderly onset of pedophilia, this latter option is unlikely. Dr. Land attempted to evade the former conclusion by guessing that Rodriguez may have committed earlier sexual acts with children which went undetected. This certainly is a possibility but there is no such evidence. It would be irresponsible for the Court to reach a conclusion as to whether a respondent has a personality disorder based on such pure speculation. Rather, the Court must base its decision on the evidence in the record and that evidence reflects that Rodriguez had not even been accused of any sexual misconduct until he was 60 years old.9
Based on this analysis, this Court finds that Dr. Land’s opinion that Rodriguez suffers from a personality disorder as defined in G.L.c. 123A, §1 is not credible and is not reliably premised on either clinical experience or empirical evidence. Dr. Kriegman’s opinion that Rodriguez does not suffer from a personality disorder, on the other hand, is credible and is premised on reliable empirical evidence. While a finding of no probable cause could rest on this finding alone, this Court will nonetheless address the third element needed to find someone a sexually dangerous person under G.L.c. 123A, §1 — that the person is likely to engage in sexual offenses if not confined to a secure facility. Pragmatically, if the evidence supported a high likelihood of sexual recidivism, this Court would revisit its finding regarding the absence of reliable evidence as to the second element.
2. Is Rodriguez Likely to Engage in Sexual Offenses If Not Confined to a Secure Facility?
There are at least three methods to predict whether a sexual offender is likely to engage in sexual offenses if not confined to a secure facility. The first is clinical judgment, “in which expert opinion is used to weigh a variety of information gained through interviews, formal testing, and offense history.” R.K. Hanson & Bussiere, M.T., Predictors of Sexual Offender Recidivism: A Meta-Analysis at 16. The second is statistical analysis, in which, through multiple regression analysis, one devises an algorithm that incorporates the variables most closely correlated with sexual recidivism to determine the statistical probability that a sexual offender with those pertinent characteristics will re-offend within a selected time frame (e.g. within five, ten, or fifteen years). Id. A third is guided clinical judgment, where the probability of sexual recidivism is determined by a combination of clinical judgment and statistical analysis, such that the clinical judgment is given a limited degree of weight along with the statistically predictive variables. Hanson, R. Karl, “What Do We Know About Sex Offender Risk Assessment,” Psychology, Public Policy and Law, at 61-62.
Clinical judgment is the predictive means anticipated by the Legislature when it enacted the 1999 legislation regarding sexually dangerous persons, because it provided that, after the probable cause hearing, the respondent must be examined and diagnosed by two qualified examiners. G.L.c. 123A, §13(a). The problem with clinical judgment is that, empirically, it has proven to have limited validity. “In general, the average predictive accuracy of professional judgment to predict sex offense recidivism is only slightly better than chance (average r=.10).” R.K. Hanson & Thornton, David, Static 99: Improving Actuarial Risk Assessmentsfor Sex Offenders at 2. Statistical risk prediction scales, developed from regression analysis and using multiple variables that have been shown to be correlated with sexual recidivism, have substantially greater predictive accuracy than clinical judgment. R.K. Hanson & Bussiere, M.T., Predictors of Sexual Offender Recidivism: A Meta-Analysis at 16. The test that best predicts sexual recidivism, though far from perfectly, is the Static-99 Sex Offender Risk Assessment scale devised by Hanson, which has achieved an r value of .33 when applied to multiple data samples. R.K. Hanson & Thornton, David, Static 99: Improving Actuarial Risk Assessments for Sex Offenders at 13.10 Indeed, the Massachusetts Sex Offender Registry Board intended to use the Static 99 scale to determine the risk of sexual offenders included within the Registry. It may be (although it has yet to be established) that guided clinical judgment can improve upon the predictive power of statistical risk prediction scales when there is specific information personal to the sexual offender that would affect the likelihood of recidivism but not show up as a variable measured in the algorithm, such as the offender’s professed willingness to re-offend. See Hanson, R. Karl, “What Do We Know About Sex Offender Risk Assessment,” Psychology, Public Policy and Law, at 61 (evaluators need to consider variables that have yet to receive full empirical validation, such as stated intention to re-offend). However, since statistics, in general, are better predictors of future sexual dangerousness than clinical judgments, caution needs to be used in adjusting a statistical prediction based on clinical judgment. See id. Pragmatically, this means that, while a factfinder [33]*33may have good reason modestly to adjust a statistical probability based on idiosyncratic clinical information unique to the sexual offender, there would need to be a compelling reason to justify a dramatic adjustment to the statistical probability.
Dr. Land had not interviewed Rodriguez and did not offer a clinical judgment regarding his risk of sexual recidivism. Rather, he reached his opinion based on his understanding of the empirical research in the field, which he recognized had been surveyed by Hanson. From this understanding, he opined that a sexual offender with multiple victims and multiple sexual offenses had abase rate of recidivism ofbetween 40-60 percent, while a sexual offender with a single victim and multiple sexual offenses had a base rate of recidivism of between 30-40 percent. He placed Rodriguez in the latter category. He then opined that Rodriguez’s refusal to participate in prison in the Sex Offender Treatment Program increased his rate of recidivism to roughly 50 percent, and that his alcohol abuse increased his risk even higher.
There are at least three problems with Dr. Land’s rough statistical analysis. First, Dr. Land did not know whether these probabilities reflected the rate of sexual recidivism (the likelihood that a sex offender would commit a sex offense in the future) or general recidivism (the likelihood that a sex offender would commit a criminal offense in the future). Only the sexual recidivism rate is relevant to determine whether someone is a sexually dangerous person. The general recidivism rate is bound to be much greater, so reliance on that probability is substantially misleading to the Court. For the 61 studies surveyed by Hanson, the rate of sexual recidivism was 13.4 percent; the rate of general recidivism was 36.3 percent, nearly three times higher. R.K. Hanson & Bussiere, M.T., Predictors of Sexual Offender Recidivism: A Meta-Analysis at 13.
Second, Dr. Land did not know what time period was reflected in these probabilities — he did not know whether the new offense occurred within 5, 10, 15, or 50 years of the original sexual offense. Obviously, the recidivism rate is going to be higher the longer the time period. Based on Hanson’s actuarial analysis, which he incorporated into the Static-99 scale, a sex offender with a specified “low medium” risk of sexual recidivism had a 12 percent probability of sexual recidivism within 5 years but a 19 percent probability within 15 years. R.K. Hanson & Thornton, David, Static 99: Improving Actuarial Risk Assessments for Sex Offenders at 12. See generally R.K. Hanson & Bussiere, M.T., Predictors of Sexual Offender Recidivism: A Meta-Analysis at 7 (“statements about recidivism rates have little meaning without specifying the definition and followup period”).
Third, and foremost, he could not identify where his understanding of the empirical evidence came from; he was not able to identify a single article he had read or specify any source for his understanding. To the extent he relied on Hanson’s survey of the empirical studies of sexual recidivism, he is mistaken. As noted earlier, Hanson’s survey of 61 studies, including9,603 child molesters, found a sexual recidivism rate over 4-5 years of 12.7 percent. R.K. Hanson & Bussiere, M.T., Predictors of Sexual Offender Recidivism: A Meta-Analysis at 13. Hanson also found that a number of characteristics possessed by Rodriguez were significantly correlated with a lower risk of sexual recidivism. “(T]he younger sexual offenders were more likely to recidivate than were the older sexual offenders.” Id. at 14. “Those who selected related child victims (incest offenders) were at lower risk than were other sexual offenders.”11 Id. “Offenders against female children were, on average, less likely to recidivate than were the other offenders (e.g., rapists, offenders against boys, exhibitionists against adult women).” Id. See also DSM-IV-TR at 571 ("The recidivism rate for individuals with Pedophilia involving a preference for males is-roughly twice that for those who prefer females”). This Court recognizes that Hanson found that prior sexual offenses were significantly correlated with a higher risk of sexual recidivism and that, if one treats Rodriguez’s indecent assault and battery against S as a prior sexual offense, this would increase his statistical likelihood of sexual recidivism, but the effect of this variable is not overwhelming (r=.19). Id. There is no way, based on the empirical evidence surveyed by Hanson, that Rodriguez’s statistical probability of sexual recidivism comes close to the probabilities opined by Dr. Land.
As noted earlier, this Court recognizes, as observed by Hanson, that “any empirical estimates of sexual offenders’ recidivism rates should be considered underestimates,” since most sexual offenses never come to official attention. Id. at 7. This Court also recognizes that presently we cannot know the extent to which these recidivism rates underestimate the actual rate. Yet, statistical analysis does permit us accurately to assess sexual offenders’ relative risk of recidivism, that is, whether a sexual offender with certain characteristics poses a higher or lower risk of reoffending than sexual offenders with different characteristics. Id. Pragmatically, the inherent limitations of the data mean that empirical statistical analysis allows only a rough estimate of the magnitude of the absolute risk of sexual recidivism, but can provide a far better estimate of the relative risk of sexual recidivism among sexual offenders. Knowing the relative risk is critically important since, given the large numbers of persons who commit sexual offenses in the Commonwealth each year, it is not realistic to civilly commit those sexual offenders with a low relative risk of offending.
If one measures Rodriguez’s risk of sexual recidivism on the Static-99 scale, the predictive tool which, according to the record in this case, despite its significant limitations, is still the best predictive tool available, one discovers that Rodriguez’s relative risk is [34]*34low-medium among sexual offenders. R.K. Hanson & Thornton, David, Static 99: Improving Actuarial Risk Assessments for Sex Offenders at 18-19, Appendix I. Static-99 uses ten risk factors which are significantly correlated with recidivism (either positively or negatively), and places a value on the presence or absence of that factor with respect to the offender. Id. The values are simply 0 or 1 for all factors but one: three to five prior sex offenses produces a value of 2, and six or more produces a value of 3. Id. Rodriguez scores a value of 0 as to eight of the ten risk factors: he has three or less sentencing dates; no convictions for non-contact sex offenses; no convictions for prior non-sexual violence;12 no unrelated victims;13 no stranger victims; no victims the same age as the offender; he was 25 years or older when assessed; and he had stable adult relationships, with both his wife and, later, S’s mother, that lasted for two years or more. Id. With the other two risk factors — prior sex offenses and index non-sexual violence (referring to violence accompanying the most recent sexual offense) — Rodriguez scores a value of 1 for one prior sex offense (the December 1995 indecent assault and battery of S) and 1 for the assault and battery on a police officer that he committed at the time of his arrest in July 1996, for a total score from all his risk factors of 2.14 A score of 2 places Rodriguez in the lower half of the “Low Medium” risk category, translating into an absolute sexual recidivism probability of 9 percent within the next five years, 13 percent within the next ten years, and 16 percent within the next fifteen years. Id. at 20; R.K. Hanson & Thornton, David, Static 99: Improving Actuarial Risk Assessments for Sex Offenders at 12. Among the sex offenders in the Static-99 sample, 57.6 percent were in a higher risk group than Rodriguez. R.K. Hanson & Thornton, David, Static 99: Improving Actuarial Risk Assessments for Sex Offenders at 12.
In view of both Rodriguez’s relative risk of sexual recidivism among sexual offenders and the magnitude of his absolute risk of sexual recidivism, no factfinder reasonably could find that he was “likely to engage in sexual offenses if not confined to a secure facility,” as required by G.L.c. 123A, §1. In short, there is a risk that Rodriguez will commit a new sexual offense and that risk is significant, but there is no credible evidence that this risk reaches the likelihood needed to justify his indefinite civil commitment.
This conclusion is made even stronger when one considers that, if Rodriguez is not confined to a secure facility, he will be on probation over the next five years, with a condition that he participate in sex offender counseling as deemed appropriate by the Probation Department.15 Therefore, to the extent that Dr. Land is concerned that Rodriguez is at greater risk because he is not receiving sex offender treatment, that concern should be alleviated by the fact that, until Rodriguez is 70 years old, he must participate in appropriate sex offender counseling or be in violation of his probation.
ORDER
For the reasons detailed above, this Court finds that there is not probable cause to believe that Rodriguez is a sexually dangerous person as defined in G.L.c. 123A, §1. Since Rodriguez has completed his term of imprisonment and is being held in custody only as a result of this petition, this Court ORDERS that Rodriguez be released from custody forthwith and commence his term of probation.