dissenting.
Mark Murrell is committed for treatment. No matter how much treatment he receives, and no matter how well Murrell responds to treatment, as long as he remains confined, he always will score the same on the statistically based risk assessment instruments the state misuses through expert testimony in this case.
Evidence of statistics-based risk assessment measures, the STATIC-99 and the MnSOST-R, is fundamentally at odds with the constitutional basis on which the state may seek civil commitment of sex offenders who are alleged to be sexually violent predators under section 632.480 et seq.
The constitutional basis for these commitments is that they are not punishment, but rather rehabilitative. Civil commitment under Missouri’s sexually violent predator law is not criminal punishment, whose objectives are retribution or deterrence. Such laws were upheld by the United States Supreme Court in Kansas v. Crane, 534 U.S. 407, 122 S.Ct. 867, 151 L.Ed.2d 856 (2002), and Kansas v. Hendricks, 521 U.S. 346, 117 S.Ct. 2072, 138 L.Ed.2d 501 (1997). The sexually violent predators’ confinement is for the purpose of holding the person until his mental abnormality no longer causes him to be a threat to others, and he is permitted to be released on a showing that he is no longer dangerous. See section 632.495; Hendricks, 521 U.S. at 363, 117 S.Ct. 2072.
In upholding Kansas’ sexually violent predator law, the Supreme Court noted that the maximum time an individual can be incapacitated is one year, 521 U.S. at 364, 117 S.Ct. 2072, which is similar to Missouri’s statutory scheme. This Court upheld Missouri’s sexually violent predator law, relying upon the “multiple opportunities for court review,” in finding that our statutory scheme is “narrowly tailored to promote the compelling interest of protecting the public from this small percentage of offenders.” In re Care and Treatment of Norton, 123 S.W.3d 170, 175 (Mo banc. 2003). As with the Kansas statute, the Missouri statute provides for annual examinations and court review to determine if the person’s “mental abnormality has improved with treatment and if the individual remains likely to engage in violent sexual acts if released.” 123 S.W.3d at 175, citing section 632.498. In these reviews, as this Court has noted, the state has the burden of proving that the person’s mental abnormality “has not improved and that it is not safe to release” the sexually violent predator. Id.
The statistically based evidence that this Court approves is directly contrary to these constitutionally-based premises. The evidence that purports to show that Murrell has a 52 per cent chance of re-offending in the next 15 years is based on statistical analysis of characteristics that never change. By approving these misused scientific measures, the Court assures that Murrell probably will never regain his liberty no matter how much treatment he receives and no matter how well he responds to it.
The STATIC-99 instrument was developed by researchers in Canada and the United Kingdom using two risk assessment instruments and was validated on a population of several thousand male sex offenders. The purpose of the STATIC-99 is to determine whether a male offender with certain characteristics is at low, moderate-low, moderate-high, or high risk of committing another sexual offense. The STATIC-99 does not say that Murrell himself has a 52 percent risk of re-offending. Rather, the STATIC-99 says that 52 percent of a group of male sex offenders who share Murrell’s “static” (i.e. unchanging) characteristics committed a new of*115fense within a certain period of time. It is, at most, a generalization about a group, not a prediction as to any individual’s future behavior.
The STATIC-99 is an instrument that is useful to sentencing judges in assessing the risk that a particular offender is in a category of persons who are more or less likely to re-offend and is perfectly appropriate at the sentencing stage. The STATIC-99 is being adopted for use by the Missouri Board of Probation and Parole in developing pre-trial sentencing assessment information for trial judges. This instrument is useful in informing the trial judge whether a particular offender shares the characteristics of those persons in a high, moderate, or low-risk group. Although the STATIC-99 does not, in fact, predict future behavior in a particular individual, a sentencing judge may find the assessment helpful in determining what kinds of controls, short of confinement, or what kind of program, in prison or in the community, might work to reduce the chance of recidivism in a particular type of offender. The sentence, as always, is based on the judge’s judgment; the STATIC-99 is at best useful in reinforcing that judgment.
The use of this instrument through expert testimony in the civil commitment context, however, is not only wrong, but harmful. The reason that the instrument is inappropriate in civil commitment proceedings is that the characteristics of a confined offender — such as Murrell — will never change.
The STATIC-99 risk assessment methodology is as follows: a one-point value is assigned to each of the following factors, none of which, except age, is subject to change for a confined man. The higher the score, the greater the statistical risk.
1. Age — a person between the ages of 18 and 25 is scored one point;
2. A person who has never lived with a lover or significant other for at least two years — one point;
3. Present convictions for nonsexual violence — one point;
4. Prior nonsexual violence offense— one point;
5. Prior sexual offenses are identified in three ranges, based upon the number of charges and the number of convictions: depending upon this criminal offense history, points are scored from zero to three;
6. The number of times a person has been sentenced for felonies on different occasions scores an additional point if the person has been sentenced four or more times for any offense;
7. Convictions for non-contact sexual offenses — one point;
8. Unrelated victims — one point;
9. Victims who are strangers to the offender — one point;
10. If the offender has male victims— one point.
When all of these points are added up, an offender is considered low risk if he has zero or one points; if he has two or three points, he is moderate to low risk; if he has three, four or five points, he is rated moderate to high risk; and if he has six or more points, the offender is considered high risk.1
What is most noteworthy about this scoring system is that, except for age, the risk score never changes once a person is *116confined. In other words, items 2 through 10 above will always be part of the confined offender’s history, and they do not change. If he is committed civilly with, for example, a score of six or more, when he comes up for evaluation for release, he will still have the same score. It might be one point lower if he passes from under age 25 to over age 25, but that is the only thing that can possibly change about this evaluation. Murrell’s score will never change; he is 46 years old. The other risk assessment instrument used by the expert in this case, the MnSOST-R risk assessment methodology developed in Minnesota on a smaller number of sexual offenders than the STATIC-99, shares the same characteristics that make the STATIC-99 inappropriate for use in civil commitment proceedings.
Under the statutory scheme for civil commitment of sexually violent predators, and the constitutional basis on which these laws are upheld, this statistical evidence is irrelevant in a legal sense, even if such evidence seems to be logically relevant.
Evidence must be both logically and legally relevant to be admissible. See, e.g., Shelton v. City of Springfield, 130 S.W.3d 30, 37 (Mo.App.2004). The evidence that these statistical assessment instruments present is that, out of large number of offenders, a certain percentage of those in the group having a particular score re-offended during some period of time. Does this have some logical relevance to those offenders in a group sharing Mur-rells’ characteristics? The risk instruments say nothing about him personally. The risk instruments merely say that Mur-rell has characteristics similar to those of the group whose risk of re-offending is a statistically determined probability. It says nothing about whether or not Murrell himself is likely to re-offend. In fact, the researchers who developed the STATIC-99 candidly admit that one of the instrument’s weaknesses is that “it demonstrates only moderate predictive accuracy.”2 What that means is that the STATIC-99 is better than flipping a coin for predicting the future behavior of individual sex offenders, but not that much better. In this context it is difficult to find any logical relevance.
If the discussion of logical relevance seems uncomfortably arcane, an examination of its legal relevance will bring us to a more comfortable setting. The question of legal relevance is particularly important here because the evidence that purports to present this in scientific or statistical terms is misleading. Legal relevance involves balancing the probative value of evidence against its prejudicial effect on the jury. If the probative value of the evidence is outweighed by its prejudicial effect, it should be excluded.
All that this statistical evidence tends to show is that Murrell is somewhat like members of a group of sex offenders who re-offended in the past. What the jury hears is likely to be quite different — that this sex offender, Murrell, has a 52 per cent likelihood of committing another sexual offense. What the jury heard in this case from the state’s out-of-state expert was as follows:
“Q Does it (the STATIC-99) give you any percentages or produce a result that comes out in percentages?”
“A The score that Mr. Murrell received is associated with a 52 percent chance of *117being reconvicted for a sex offense over a 15-year period of time.”3
What jury is going to vote to release him from confinement with that “scientific” prediction?
Because of its inherently misleading nature in the civil commitment context, this statistic is not evidence that assists the finder of fact, under section 490.065.1, but in fact is evidence that causes the jury to believe that there is some probative value to it. The principal opinion relies on the expert’s testimony — that these risk assessment instruments are relied upon by experts in the field — to make the results admissible under section 490.065.3.4 A court’s duty under this statute is to make sure the “facts or data in a particular case upon which an expert bases an opinion” are “of a type reasonably relied upon by experts in the field ...” and are “otherwise reasonably reliable.” (Emphasis added.) In my view, these risk assessments should be excluded if they are logically or legally irrelevant regardless of what the expert says about their use by other experts in the field because the use of irrelevant facts or data is inherently unreasonable. Nor are these facts and data “otherwise reasonably reliable,” as I have discussed.
Under this expert testimony, it is asserted that Murrell currently has a 52 per cent chance of re-offending within 15 years. More accurately stated, 52 percent of a group of untreated sex offenders who shared Murrell’s characteristics re-offended. On this basis the state says Murrell should be committed civilly.
At his next review, however, the same evidence can be adduced, under the principal opinion’s analysis. In two years, three years, and so on, upon each annual review, he will still have a 52 per cent change of re-offending within 15 years, because Mur-rell’s “static” characteristics will not have changed. This is true no matter how much treatment Murrell has received, no matter what the results.
The treatment that Murrell will be afforded by this civil commitment is designed to train him to curb his impulses. With treatment, he may become adequately trained to control his impulses, but using these statistical measures, he will still have a “52 per cent risk” of re-offending within 15 years. Does anyone remember the Soviets’ misuse of their mental health system for incarcerating enemies of the state? Does this seem at all similar?
*118The use of the STATIC-99 and the MnSOST-R methodologies in the civil commitment proceeding is a scientifically-based means to justify confining Murrell for the rest of his life regardless of the outcome of his treatment. The old Soviet Union is gone, but is its legacy of misusing its mental health system alive and well in this country?
I would grant Murrell a new trial with this statistically based evidence excluded. I respectfully dissent.
. Andrew Harris, Amy Phenix, R. Karl Hanson, & David Thornton, "STATIC-99 Coding Rules — Revised—2003.” Available online from the Solicitor General of Canada: www. sgc.gc.ca: 1) Make language choice 2) Click on "Corrections” 3) Click on "Corrections Reports” 4) Click on "2003.”
. Harris, et al. p. 3. The ROC (Receiver Operating Characteristic) curve is .71; this is a statistical way of expressing predictivhy. A ROC close to 1.0 would be an excellent rating; a .5 is about the same as flipping a coin.
. It is difficult to see how the quoted testimony does not violate the principal opinion’s rationale, stated in its penultimate paragraph:
Moreover, the Court does not hold that an expert can argue based on the Static 99 that the particular offender has a 52% chance of reoffending. Rather, the Court holds the expert may base his or her opinion on actuaríais because they constitute facts or data of a type reasonably relied upon and are otherwise reasonably reliable. The expert may testify that according to actuaríais a certain percentage of people with characteristics like Murrell’s do reof-fend; that data can be part of the assessment as to whether Murrell is more likely than not to reoffend because although actu-arials are not determinative, they are relevant to that determination.
Although the expert’s statement about the STATIC-99 is couched in expert-speak ("is associated with”) it is difficult to believe that any lay person on a jury would understand this to mean something other than that Mur-rell has a 52 percent risk of re-offending.
. Section 490.065.3 provides: "The facts or data in a particular case upon which an expert bases an opinion or inference may be those perceived by or made known to him at or before the hearing and must be of a type reasonably relied upon by experts in the field in forming opinions or inferences upon the subject and must be otherwise reasonably reliable.”