¶1 Robert Bergen appeals the trial court’s order denying him a less restrictive alternative (LRA) placement as a sexually violent predator (SVP). Bergen contends that by allowing the State to defeat a proposed LRA by showing that it is not in his “best interests,” RCW 71.09.090 violates his right to due process because it is not narrowly tailored to justify commitment of violent sex offenders. He also challenges as unconstitutionally vague the statutory requirements that an LRA be in the SVP’s “best interests” and “adequately protect the community.” He contends the trial court erred by failing to give the jury his proposed instruction defining “adequate community safety.” Finally, he asserts the trial court erred by allowing testimony that he participated in annual reviews because it encouraged the jury to speculate that the previous reviews did not support conditional release. We hold the statute does not deny Bergen’s right to due process and is not unconstitutionally vague. The stated legislative intent of the SVP statute indicates that the “best interests” standard applies to the mental health treatment needs of a violent sex offender, a standard which must be met to justify release to an LRA. And the terms “best interests” and “adequate community safety” can be understood by persons of common intelligence and reasonably applied within the statute’s intent. Finally, evidence of Bergen’s annual reviews was relevant foundation evidence. We therefore affirm.

*521FACTS

¶2 On November 2, 2001, Robert Bergen was found to be an SVP and ordered committed to the Special Commitment Center (SCC). He was 74 years old at the time and was anticipating release on a 1997 conviction for communicating with a minor for immoral purposes. Bergen’s sex offense history spans 50 years, from 1947 through 1997, and includes convictions for indecent liberties, contributing to the delinquency of a minor, second degree assault, and communicating with a minor for immoral purposes, all of which involved minors, both male and female. The most recent offense occurred at his home, approximately 150 feet from an elementary school. At that time, witnesses reported Bergen talked to children passing by his home and parked his car by the school to offer to fix children’s bicycles.

¶3 In 2005, Bergen retained an independent forensic psychologist, Dr. Robert Prentky, to evaluate a plan for conditional release to an LRA placement. Based on the results of this evaluation, Bergen petitioned for an LRA under RCW 71.09.090. In support of the petition, Dr. Prentky advised that given Bergen’s advanced age, an LRA would be in his best interest and the proposed plan would adequately protect the community. On April 27, 2006, the trial court entered an order finding cause to schedule a jury trial to determine whether Bergen should be granted conditional release to an LRA.

¶4 At trial, Bergen presented his proposed conditional release plan, which returned him to the home where he committed the most recent offense. He proposed to reside there with his wife, follow the conditions of his sentence imposed on the 1997 conviction, and attend sex offender treatment. His plan also provided that he would register as a sex offender, agree to Global Positioning System (GPS) surveillance, not drive, not use the Internet, and comply with all conditions imposed by his supervising corrections officer.

*522¶5 Prentky also testified and opined that the release plan served Bergen’s best interests because it would allow him to live out the remainder of his life with his wife under less restrictive circumstances. He also described the plan as “close to water tight” and testified that it would provide more than adequate community safety. Prentky did not interview Bergen but testified that personal contact with Bergen was unnecessary because he was asked to address the “fairly narrow question” of whether the LRA was appropriate.

¶6 The State presented testimony from Dr. Paul Spizman, who conducted three statutorily required annual review evaluations1 of Bergen. Spizman interviewed Bergen, his wife, his nurse practitioner, and his proposed treatment provider. Spizman opined that Bergen suffers from pedophilia and antisocial personality disorder and that he would continue to be a high risk to reoffend if released under the proposed LRA.

¶7 Dr. Jonathon Allison, a forensic evaluator at the SCC, also testified for the State about an interview he conducted with Bergen in 2006 during an annual review. During that interview, Bergen said that he refused to participate in treatment at the SCC and that treatment was an indication of weakness. He also told Allison that he never hurt anyone “except to knock them unconscious,” described how he could do that with a “choke hold,” and said that his hand “was classified as a lethal weapon.” He denied the 1996 offense and said that if he was in a car and saw the victim, he would press his foot on the accelerator. He also told Allison, “ T would have been better off if I had killed the boy in my residence. My only mistake was to let him go.’ ”

¶8 The State also called Randy Green, the treatment provider Bergen suggested in his release plan. Green testified that he interviewed Bergen in December 2005 and that he was willing to treat Bergen if he was conditionally released. He also testified that the LRA was “as good as it’s *523going to get,” but that he had concerns about it. He questioned the effectiveness of GPS surveillance to prevent offenses of the type Bergen most recently committed and the appropriateness of his wife as a chaperone given her age and denial of his history. He also noted Bergen’s history of denial and remorselessness. He further testified that given Bergen’s past treatment failures, it was unlikely that the proposed treatment would have a positive effect on him.

¶9 Additional witnesses for the State included Randall Griffith, Bergen’s nurse practitioner at the SCC; Joseph Beard, a detective who oversees and conducts registration of sex offenders in Snohomish County; and Tela Wilson, a community corrections officer. Griffith testified that aside from mild arthritis, Bergen was “very fit for his age” and “very mobile.” Beard testified that the Department of Corrections asked him to investigate Bergen’s proposed LRA, including the proposed release address. He said a day care, elementary school, soccer field, and parks were in close proximity to Bergen’s home, and school bus stops were visible from his home. Wilson testified that GPS surveillance monitors whether an offender is within 75 to 200 feet of the home and alerts a corrections officer once the offender leaves that vicinity.

¶10 The trial court instructed the jury that the State had to prove beyond a reasonable doubt that Bergen’s proposed LRA was not in his best interests or did not include conditions that would adequately protect the community.2 The court refused to give Bergen’s proposed instruction defining “adequate community safety.” The jury returned a verdict finding that the State had proved both factors beyond a reasonable doubt.

I. “Best Interests” Standard

A. Due Process

¶11 Bergen first argues that the statutory provision that allows the State to defeat a LRA based on proof that it is not *524in the offender’s best interest violates his right to due process. He argues that the “best interests” standard in ROW 71.09.090 does not relate to dangerousness and mental illness concerns that justify involuntary commitment. Thus, he contends, it is not narrowly tailored to serve the compelling state interest in treating sex predators and protecting society from their actions.

f 12 A statute is presumed constitutional, and the party challenging it bears the burden of proving it is unconstitutional beyond a reasonable doubt.3 The due process clause of the constitution requires that a person shall not be deprived of life, liberty, or property without due process of law.4 When a state’s laws impinge on fundamental rights, they must further compelling state interests and be narrowly drawn to serve those interests.5 The threshold question in any due process challenge is whether there has been a deprivation of a protected interest in life, liberty, or property.6 When there is no alleged violation of a fundamental right, the challenged state action is not subject to strict scrutiny but need only be rationally related to a legitimate government interest.7

¶13 Bergen asserts that he has a fundamental liberty interest in his conditional release because “[i]nvoluntary civil commitment and indefinite detention are serious infringements of an individual’s liberty interest.”8 Bergen cites case law involving due process challenges to the initial *525SVP commitment, not to a postcommitment petition for an LRA, which is at issue here.9 Thus, we must first determine whether Bergen has a liberty interest in a petition for an LRA once he has already been committed as an SVP. In this analysis, it is important to recognize that Bergen does not contend he is no longer an SVP. He seeks release into the community as an adjudicated SVP.

¶14 Liberty interests may arise from either of two sources: the due process clause and state laws.10 The due process clause does not, of its own force, create a liberty interest when an inmate seeks release before serving the full maximum sentence.11 Similarly, the due process clause does not create a liberty interest when an SVP seeks release before the court has determined that he or she is no longer likely to reoffend or that he or she is entitled to conditional release to an LRA.12

¶15 But state statutes or regulations can create due process liberty interests where none would have otherwise existed.13 “By enacting a law that places substantive limits on official decisionmaking, the State can create an expectation that the law will be followed, and this expectation can rise to the level of a protected liberty interest.”14 Thus, laws that dictate a particular outcome based on particular facts can create liberty interests, but laws granting a significant degree of discretion cannot.15

¶16 In In re Personal Restraint of Cashaw, the court held that certain procedural regulations for parole hearings did *526not create a liberty interest.16 Because an inmate cannot be released unless the Indeterminate Sentence Review Board determines that the inmate has been rehabilitated, the court concluded that the decisions on an inmate’s parolability are not guided by “ ‘substantive predicates’ ” and “ ‘specific directives’ ” from which “ ‘a particular outcome must follow.’ ”17 Rather they were “ ‘subjective appraisals’ ” of the degree to which an inmate has been rehabilitated and a “ ‘discretionary assessment of a multiplicity of imponderables.’ ”18

¶17 Similarly, in In re Detention of Enright, this court held that an SVP does not have a liberty interest in a hearing on his classification as a level III sex offender.19 Citing Cashaw, the court held that because risk assessment classifications involve “subjective appraisals” by a committee that assigns risk levels and local law enforcement, both of which are vested with significant discretion in making that determination, the sex offender registration and disclosure statutes did not create a substantive liberty interest in the risk assessment classification.20

f 18 Here, as in Enright, the due process clause does not create a liberty interest in a conditional release to an LRA because an SVP offender does not have a liberty interest in being released before a court determines that the SVP is entitled to such a release. But RCW 71.09.090, the statute that provides for an LRA, contains “substantive predicates” and “specific directives” from which “a particular outcome must follow.” The statute provides that an SVP who petitions for an LRA is entitled to a show cause hearing in which the court determines whether probable cause *527exists to warrant a hearing on whether the conditional release to an LRA is appropriate.21 If the court determines probable cause exists, the SVP is entitled to a jury trial in which the State may defeat the LRA only by proof beyond a reasonable doubt that the LRA either is not in the offender’s best interests or does not adequately protect the community.22

¶19 Thus, unlike the parole procedures in Cashaw or the classification procedures in Enright, this determination is not based on discretionary assessments or subjective appraisals by nonjudicial entities or law enforcement, but involves a judicial determination of probable cause and requires the State to affirmatively defeat the LRA by proof beyond a reasonable doubt of one of two statutory factors. The statute further provides that at this hearing, the SVP shall be entitled “to the benefit of all constitutional protections that were afforded to the person at the initial commitment proceeding.”23 Thus, the statutory provisions that allow an SVP to petition for an LRA dictate a particular outcome based on particular facts and therefore create a liberty interest in a conditional release to an LRA. Consequently, we will apply strict scrutiny and determine whether the statutory procedures are narrowly tailored to serve a compelling state interest.

¶20 In reviewing substantive due process challenges to the SVP statute, our courts have recognized that “the State has a compelling interest both in treating sex predators and protecting society from their actions.”24 But the State must prove that a person is both mentally ill and dangerous to justify civil commitment under the due process clause of the constitution.25 Here, the State met that burden when Bergen was adjudicated an SVP. As the State *528points out, he does not challenge that finding or seek release but only seeks an alternative placement as an SVP. Thus, his continued commitment is still supported by findings of mental illness and dangerousness and his unchallenged status as an SVP. The LRA determination is a separate inquiry and is focused on whether the SVP — who has already been found to be dangerous and mentally ill — should be transferred to an LRA that will continue to serve the statutory objectives of treating the SVP and keeping the community safe.

¶21 The stated legislative intent of RCW 71.09.090 includes the following findings about the appropriateness of LRAs:

[C]ivil commitment pursuant to chapter 71.09 RCW address [es] the “very long-term” needs of the sexually violent predator population for treatment and the equally long-term needs of the community for protection from these offenders. The legislature finds that the mental abnormalities and personality disorders that make a person subject to commitment under chapter 71.09 RCW are severe and chronic and do not remit due solely to advancing age or changes in other demographic factors.

... [A] mere advance in age or a change in gender or some other demographic factor after the time of commitment does not merit a new trial proceeding under RCW 71.09.090. [A] new trial ordered under [such] circumstances . . . subverts the statutory focus on treatment and reduces community safety by removing all incentive for successful treatment participation in favor of passive aging and distracting committed persons from fully engaging in sex offender treatment.

. . . The legislature has, under the guidance of the federal court, provided avenues through which committed persons who successfully progress in treatment will be supported by the state in a conditional release to a less restrictive alternative that is in the best interest of the committed person and provides adequate safeguards to the community and is the appropriate next step in the person’s treatment.[26]

*529Thus, the “best interests” standard accounts for the inherent dangerousness of SVPs and their unique, extended treatment needs: it relates to the SVP’s successful treatment, ensuring that the LRA does not remove “incentive for successful treatment participation” or “distract [ ] committed persons from fully engaging in sex offender treatment” and is the “appropriate next step in the person’s treatment.” The legislative findings are also clear that other factors such as advanced age, change in gender, or other demographics alone are not enough to warrant consideration of an LRA. The “best interests” standard is directly related to the SVP’s dangerousness and mental illness and is narrowly tailored to serve the State’s compelling interest in appropriately treating dangerous sex offenders.

¶22 Bergen cites O’Connor v. Donaldson27 for the proposition that the best interests standard results in impermissible involuntary confinement of the mentally ill “ ‘merely to ensure them a living standard superior to that they enjoy in the private community.’ ”28 But O’Connor involved the involuntary commitment of a nondangerous mentally ill person who was not receiving treatment and held that the State cannot constitutionally confine an individual who is dangerous to no one and can live safely in freedom.29 It therefore does not apply here, where the committed individual has already been found to be a danger to the community and does not challenge that finding.

¶23 Bergen also argues that because a treatment plan is a prerequisite for any conditional release plan, an additional “best interests” element does not ensure that an SVP receives treatment. But the “best interests” determination does not simply require a treatment program; it ensures that the proposed treatment plan is an appropriate one and will further “successful treatment participation.” Bergen further contends the “best interests” standard also results *530in the “constitutional problem” of requiring a person confined as an SVP to meet a more stringent standard to secure his release than the State had to establish for his initial confinement.30 But here, Bergen is not seeking to secure his release and challenge his SVP status; he simply seeks an alternative placement as an SVP, which, as discussed above, involves a separate inquiry and a different showing.

B. Vagueness

¶24 Bergen also argues that the term “best interests” is “so subjective that submitting instructions on this element violated procedural due process.”31 The due process vagueness doctrine under the Fourteenth Amendment to the United States Constitution and article I, section 3 of our state constitution requires that (1) the public is provided with adequate notice of what conduct is proscribed and (2) the public is protected from arbitrary enforcement.32 A statute is unconstitutionally vague under the Fourteenth Amendment if it is “framed in terms so vague that persons of common intelligence must necessarily guess at its meaning and differ as to its application.”33 Determining whether a statute sufficiently defines an offense “does not demand impossible standards of specificity or absolute agreement.”34 For a statute to be unconstitutional, its terms must be “ ‘so loose and obscure that they cannot be clearly applied in any context.’ ”35 Unless First Amendment rights are involved, a vagueness challenge to *531statutory terms may be considered only “as applied.”36 Thus, the court must evaluate the statute as applied to the actual conduct of the party challenging the statute.37

¶25 Here, Bergen’s vagueness challenge is based on the trial court’s failure to define “best interests” in the jury instructions. Bergen argues that the question of what might be in his best interest was “so amorphous and subjective,” speculating that some jurors might have believed continued confinement was in his best interest because he was not at risk to reoffend against a minor, while others might have believed that community notification requirements might pose threats to his safety if released to his proposed LRA placement, and still others might have believed that continued confinement would be in his best interest because he was unlikely to succeed in his LRA placement and would be more angry when returned to the SCC than if he were never released at all.

¶26 But all of these scenarios fall reasonably within the “best interests” determination contemplated by the statute. As discussed above, the legislative intent indicates that the “best interests” standard relates to the appropriateness of treatment for an SVP who has been found to be both mentally ill and dangerous, and all of these examples implicate treatment concerns of a mentally ill and dangerous sex offender. And when read in context of the legislative intent and his proposed LRA, which included a treatment plan, an ordinary person would understand that determining whether an LRA is in his “best interests” involves considering whether it would adequately serve his treatment needs as an SVP. Bergen fails to establish that the term “best interests” is “ ‘so loose and obscure that [it] cannot be clearly applied in any context.’ ”38 We therefore reject his vagueness challenge.

¶27 Bergen further contends that even if the “best interests” standard is not vague, it became superfluous to the *532“adequate community safety” element when the State argued that it was in his best interests that he not reoffend. But as the State points out, reoffense would not only result in jeopardizing community safety, but would also result in violating his treatment conditions, terminating his treatment, possible criminal charges and prosecution, and a potential return to prison. All of these consequences relate to the best interests determination and do not necessarily implicate community safety.

II. “Adequate Community Safety”

¶28 Bergen next contends that the term “adequate community safety” is unconstitutionally vague. He also argues that the trial court erred by refusing his proposed jury instruction that defined “adequate community safety” as a risk of reoffense less than 50 percent.

¶29 The language that is the subject of a vagueness challenge does not require exact specificity; rather it “must be susceptible to understanding by persons of ordinary intelligence.”39 Our state Supreme Court rejected a similar vagueness challenge to the language “safe to be at large,” which was later replaced by the current language “adequately protect the community” in a 2001 amendment to chapter 71.09 RCW.40 The court held that “[a]mple standards are present to guide the exercise of discretion and to provide notice to potential detainees of prohibited conduct.”41 As the State contends, “adequate community safety” is even more specific than “safe to be at large” and clearly conveys the idea that an LRA must adequately address community safety concerns. Each of these words is commonly used, and persons of common intelligence do not need to guess at their meaning. As such, they also provide adequate notice of prohibited conduct.

*533¶30 Bergen argues that the jury could envision a number of different scenarios that would defeat this condition. But the jury is simply deciding whether the facts presented (i.e., the proposed LRA and its effect on community safety) have met a particular standard (i.e., adequate community safety), which is precisely what juries do. That the jury makes this determination does not somehow render the term vague, as Bergen seems to suggest. Bergen fails to show that the term is so vague that it violates due process.

¶31 Bergen further argues that the trial court erred by refusing his definition of “adequate community safety.” The specific language of jury instructions is a matter left to the trial court’s discretion.42 A trial court properly refuses an instruction that does not correctly state the law.43

¶32 Bergen asserts that because an SVP finding must be based on proof that the probability of reoffense exceeds 50 percent, the court should have defined the “adequate community safety” element of the LRA determination to mean that the LRA placement lowers the offender’s risk to below 50 percent. But as the State correctly contends, the question of recidivism risk is an issue when the SVP contests the commitment criteria and seeks an unconditional release, which is not the case here. Rather, Bergen seeks only a conditional release and does not challenge the finding that he meets the commitment criteria, including the fact that he is more likely than not to reoffend if released. Thus, the “adequate community safety” determination necessarily assumes that Bergen is likely to reoffend and the question then becomes whether the proposed LRA will prevent an otherwise-likely offense if he is released. The focus of this determination is therefore on the plan, not the person, and it would have been error for the trial court to instruct the jury that the “adequate community safety” element related *534to Bergen’s risk of reoffense rather than the sufficiency of the proposed LRA.

¶33 Nor do principles of statutory construction support the proposed instruction, as Bergen contends. Rules of statutory construction require courts to give effect to the legislature’s intent and purpose.44 In doing so, courts must first look to the statute’s plain meaning.45 If the statute does not define a term, courts give it its ordinary meaning.46 “ ‘[T]he court should assume that the legislature means exactly what it says. Plain words do not require construction.’ ”47 As the State notes, “Neither the phrase ‘ “adequate to protect the community” ’ nor any word in that phrase is further defined in RCW 71.09.” Thus, it should be given its ordinary meaning, which is not the definition Bergen proposed.

III. Evidence of Annual Reviews

¶34 Finally, Bergen contends the trial court abused its discretion by denying his motion to exclude evidence of the annual reviews conducted during his time at the SCC. He argues that the witnesses’ references to “annual reviews” allowed the jurors to infer that he had been previously deemed unsuitable for conditional release to an LRA and to assume that if they denied his LRA petition he could petition again the following year. Thus, Bergen asserts, this evidence encouraged the jury to improperly consider matters outside the record.

¶35 We review the trial court’s ruling on the admissibility of evidence for an abuse of discretion.48 The *535trial court has broad discretion in balancing the probative value of evidence with its potentially prejudicial impact.49 Comments calculated to encourage the jury to render a verdict based on facts not in evidence is improper.50

¶36 Bergen moved in limine to exclude evidence of his annual reviews, arguing that it was not relevant to whether his proposed LRA is in his best interests or adequately protects the community and that it was prejudicial because it allowed the jury to speculate that previous reviews did not support a conditional release. The trial court denied the motion, ruling that the State should not be prohibited from mentioning that the evaluations are a result of annual reviews. The court stated that it was not persuaded by Bergen’s concerns about the jury speculating that he had been denied an LRA in the past and suggested that concern could be addressed by Bergen’s testimony that he has never petitioned for release before.

¶37 At trial, State experts Dr. Allison and Dr. Spizman both testified that they did annual reviews of Bergen at the SCC. Allison testified that he did a review of Bergen in August 2006, and Bergen made some statements during the review that demonstrated that he was not an appropriate candidate for conditional release. Spizman testified that he conducted three annual reviews of Bergen, the last of which was completed March 9, 2006. Spizman testified that, to his knowledge, Bergen had never “productively engaged in a treatment program” or “substantially learned to control his risk factor.”

¶38 Thus, as the State argues, the references to the annual reviews provided the basis for the experts’ opinions on the appropriateness of the proposed LRA and were properly admitted as relevant evidence. Bergen fails to establish that any prejudice outweighed this relevance. *536While he argues that the jury could have inferred that he was already denied an LRA based on previous annual reviews, this argument is speculative and could have been addressed by evidence that this was the first time he sought an LRA, which Bergen chose not to present. As the State points out, Bergen’s expert was actually the first witness to refer to an annual review when he mentioned that he looked at the annual reviews before forming an opinion about the appropriateness of the proposed LRA.51 Bergen did not object or move to strike this testimony. The trial court did not err by allowing the witnesses to refer to Bergen’s annual reviews.

¶39 We affirm the order denying conditional release to an LRA.

Review denied at 165 Wn.2d 1041 (2009).