OPINION

BEA, Circuit Judge:

In California, voters have the power to change criminal-sentencing law at the ballot box. They can amend statutes and the state constitution. In 1988 and again in 2008, the voters exercised this power through the passage of Proposition 89 and then Proposition 9. Proposition 89 amended the California Constitution to vest in the Governor constitutional authority to reverse, affirm, or modify grants of parole as to inmates convicted of murder. Such authority had previously been vested solely in the Board of Parole Hearings. Proposition 9 amended the California Penal Code to increase the default period of time after which a prisoner would be scheduled for a parole hearing, after the denial of parole.

No party to this action challenges the authority of voters to make such changes. However, California inmates who were sentenced to life terms with the possibility of parole for murders committed before the passage of the two Propositions, led by Richard Gilman, contend that applying the Propositions to them creates a significant risk that their periods of incarceration will be longer than they would have been before the passage of the Propositions. If the application of either Proposition creates a significant risk of a longer period of incarceration, the Proposition violates the Ex Post Facto Clause of the Federal Constitution. Gilman and two classes of similarly situated plaintiffs sued under 42 U.S.C. § 1983 to enjoin the application of Propositions 89 and 9 as to them.

After a bench trial, the district court found in favor of the plaintiffs. As to the class members who were convicted of crimes committed before the passage of Proposition 89, the district court enjoined the Governor from imposing a longer sentence than that required by application of the same factors the Board of Parole Hearings is required to consider. The district court further ordered the Board of Parole Hearings, after denying a class member parole, to schedule that inmate's next parole hearing according to the deferral periods in place before the passage of Proposition 9. We reverse.

I. Facts and Procedural History

Until 1988, the California Board of Parole Hearings ("Board") had the exclusive power to make parole decisions. In 1988, California voters passed Proposition 89, which amended the California Constitution to grant the Governor the authority to affirm, modify, or reverse decisions of the Board with respect to inmates convicted of murder.1

In 2008, through another ballot initiative, which did not affect Proposition 89, California voters changed the parole scheme again, this time by statutory amendment, in Proposition 9.2 Before the passage of Proposition 9, prisoners sentenced to life with the possibility of parole received an annual parole-suitability hearing by default. After denying such a prisoner parole, if the Board determined that it was not reasonable to expect that the prisoner would be granted parole within a year, the Board could schedule the prisoner's next parole hearing up to five years later for murderers and up to two years later for non-murderers. Following the passage of Proposition 9, after denying such a prisoner parole, the Board may schedule his next parole hearing fifteen, ten, seven, five, or three years later (the "deferral periods").

Notwithstanding these deferral periods, Proposition 9 allows an inmate to request that the Board advance the date of his next parole hearing. To do so, an inmate submits a petition to advance ("PTA") setting forth "the change in circumstances or new information that establishes a reasonable likelihood that consideration of the public safety does not require the additional period of incarceration of the inmate." Cal.Penal Code § 3041.5(d)(1). The Board has sole discretion to grant or deny a PTA; it may also advance an inmate's next parole hearing sua sponte. Id. § 3041.5(b)(4), (d)(2). If the Board denies the inmate's PTA, the inmate may not submit another PTA for three years. Id. § 3041.5(d)(3).

In 2005, Gilman and other California inmates convicted of murders committed before November 2, 1988, sued the State under 42 U.S.C. § 1983. Gilman alleged that Proposition 89 retroactively increased the punishments of class members, in violation of the Ex Post Facto Clause, and sought to enjoin the enforcement of Proposition 89. In 2009, Gilman amended and supplemented his complaint to allege that Proposition 9 also violated the Ex Post Facto Clause. To that end, he added a subclass composed of inmates who were convicted of an offense committed on or after November 8, 1988, the date of Proposition 89's passage, but before November 4, 2008, the date of Proposition 9's passage.3

Gilman moved for a preliminary injunction to bar enforcement of Proposition 9 based on the allegations that it violated the Ex Post Facto Clause. The district court ruled that Gilman was likely to succeed on the merits of his claim and granted the motion. The State filed an interlocutory appeal and, in a published opinion, we reversed, "[b]ecause on the current record Proposition 9 does not create a significant risk of prolonging [Gilman's] incarceration on any of the theories [he] assert[s], [and] [Gilman] ha[s] not established that [he is] likely to succeed on the merits of [his] ex post facto claim." Gilman v. Schwarzenegger (Gilman I), 638 F.3d 1101, 1111 (9th Cir.2011).

At a bench trial, as to the Ex Post Facto Clause claim against Proposition 89, Gilman proffered evidence showing that between 1991 and 2010 the Governor reversed more than 70% of the Board's decisions granting parole to prisoners with murder convictions. The district court found that most such reversals were related to prisoners "who were already beyond their `life terms,' so that but for Proposition 89 and the Governor's reversal, they would have been released already."4

Based on this evidence, the district court found that Proposition 89 "was passed in order to lengthen the amount of time class members would spend in prison by creating a new mechanism for withholding parole, namely, the governor's veto" and, "[t]rue to the law's intentions, California governors have used [Proposition 89] to withdraw the possibility of parole from most class members." The district court thus held that Proposition 89, as implemented by California Governors, is a "plain violation of the ex post facto clause as to those inmates whose crimes were committed before Proposition 89." The district court ordered the Governor to "refrain from imposing longer sentences on class members than are called for by application of the same factors the Board is required to consider, as provided for by Proposition 89."

As to Proposition 9, the district court made findings regarding the comparative frequency of parole hearings, and comparative rates of parole grants, before and after the passage of Proposition 9. To do this, the district court relied on the experience of inmates who had been involved in state litigation, In re Rutherford, No. SC135399A (Cal.Super.Ct., Marin Cty., filed May 26, 2004). In Rutherford, petitioner Jerry Rutherford was denied parole in 2003 and was scheduled for a hearing the next year, under the pre-Proposition 9 statute. The Board did not provide the required hearing within the year. Rutherford filed a petition in habeas corpus, in California state court, to challenge the hearing delay. The state court certified a class of "prisoners serving indeterminate terms of life with the possibility of parole who have approached or exceeded their minimum eligible parole dates without receiving their parole hearings within the time required." After class certification, the Board stipulated that it was not providing timely parole hearings. The Rutherford class and the State then agreed to a remedial plan to conduct the hearings. Some Rutherford class members had not received the hearings to which they were entitled under the plan when Proposition 9 became effective. Those class members moved for a preliminary injunction to enjoin the Board from applying Proposition 9 to them. The application for a preliminary injunction was settled by stipulation, under which some prisoners who had already had hearings (after Proposition 9 but before entry of the stipulation), had been denied parole, and had their deferral periods calculated under Proposition 9, were entitled to adjustments of their deferral periods to conform to the pre-Proposition 9 deferral periods. Those prisoners who had not received hearings stipulated to specific deferral periods or to hearings conducted under the pre-Proposition 9 statute.

The district court then made findings regarding the efficacy of the PTA process, as "the availability of advance hearings is relevant to whether the changes in the frequency of parole hearings create a significant risk that prisoners will receive a greater punishment." Gilman I, 638 F.3d at 1108. The State argued that its PTA process mitigated any risk that sentences would be unconstitutionally lengthened. The district court disagreed.

The district court found that, when he reviews a PTA, a decisionmaker designated by the Board5 analyzes the petitioner's asserted change in circumstances or new information without reference to whether a prisoner has made "a move from unsuitability to suitability" and that some prisoners who had become suitable would not be able to advance their hearings by filing a PTA. That is, the district court concluded that a PTA's contention that a prisoner is now suitable, though he was unsuitable before, is not in itself considered a change in circumstances. The district court found that this interpretation meant that some prisoners who would have been found suitable for parole at their next suitability hearing under the pre-Proposition 9 version of Cal.Penal Code § 3041.5(d)(1) were not able to advance their hearing to less than the three-year deferral period, because their suitability was not deemed a changed circumstance or new information, and so an advance hearing was not granted.

The district court pointed to the example of one prisoner, M. Brodheim, who filed a federal habeas petition on due-process grounds after the Board had denied him parole; he claimed that the Board lacked "some evidence" that he was unsuitable for parole.6 The district court agreed and entered a conditional writ of habeas corpus, to issue unless Brodheim were given a parole hearing. The State appealed and, while the appeal was pending, held a Board hearing, at which the Board found Brodheim suitable for parole. We subsequently reversed the judgment which granted the conditional writ on the ground that the district court erred in its finding that the Board lacked "some evidence" of Brodheim's unsuitability for parole. The Board then vacated its suitability finding and set Brodheim for a three-year deferral under Proposition 9. Brodheim filed a PTA, citing his suitability for parole as found by the Board. The Board denied his PTA on the ground that there was no evidence of the requisite changed circumstances or new information, although it had found him suitable for parole. The district court found Brodheim's situation to be evidence that, in some cases, the Board does not "upon request, schedule advance hearings for prisoners who become suitable." Gilman I, 638 F.3d at 1109.

The district court also found that the PTA process did not prevent a significant risk of lengthened incarceration for prisoners whose parole determinations hinged on psychiatric evaluations known as the Comprehensive Risk Assessment ("CRA") (administered every five years) and the Subsequent Risk Assessment ("SRA") (administered before a parole-suitability hearing) because an SRA could be ordered only in conjunction with a hearing, not a PTA. Therefore, the district court reasoned, a prisoner could not show changed circumstances through his risk assessment, as one would not be conducted in connection with the PTA.

Similarly, the district court found that the PTA process would not advance hearings for some prisoners whose PTAs were written in Spanish, because the decisionmakers assigned to review the PTAs could not review documents in Spanish and "could not determine whether the standard had been met until the documents were translated." The district court pointed to one instance in which a PTA was denied on this ground.

The district court concluded that "the PTA process is not sufficient to protect inmates from the ex post facto problems inherent in Proposition 9." Based on its findings, the district court ordered the Board to apply the pre-Proposition 9 version of Cal.Penal Code § 3041.5 to class members. The State appealed the district court's decision.

II. Standard of Review

We review the district court's legal conclusions de novo, its factual findings for clear error, and the scope of the injunctions for abuse of discretion. Armstrong v. Brown, 768 F.3d 975, 979 (9th Cir.2014).

III. Analysis

A change in law violates the Ex Post Facto Clause of the Federal Constitution when it "inflicts a greater punishment[] than the law annexed to the crime, when committed." Peugh v. United States, ___ U.S. ___, 133 S.Ct. 2072, 2078, 186 L.Ed.2d 84 (2013) (internal quotation marks omitted). In Gilman I, we set forth the relevant inquiry:

638 F.3d at 1106 (alterations in original) (quoting Garner v. Jones, 529 U.S. 244, 255, 120 S.Ct. 1362, 146 L.Ed.2d 236 (2000)). We already rejected Gilman's facial challenge to Proposition 9 in Gilman I, and now Gilman brings an as-applied Ex Post Facto Clause challenge based on "evidence drawn from Proposition 9's practical implementation." Id.

Although we stated that Gilman must show that the retroactive application of Proposition 9 "will result in a longer period of incarceration," we think it is enough for Gilman to prove that Proposition 9 "created a significant risk of increasing his punishment." Garner, 529 U.S. at 255, 120 S.Ct. 1362 (emphasis added). Garner, on which we relied in Gilman I, used both forms of the test, but ultimately concluded that "respondent must show that as applied to his own sentence the law created a significant risk of increasing his punishment. This remains the issue in the case." Id. (emphasis added).7 Moreover, any doubt we may have had on this point was dispelled by the Supreme Court's decision in Peugh, which was decided after Gilman I, and reiterated that "[o]ur ex post facto cases . . . have focused on whether a change in law creates a `significant risk' of a higher sentence." 133 S.Ct. at 2088.8 That is the correct standard, which we now apply to Gilman's challenges to Propositions 89 and 9.

A. Proposition 89

We have already addressed the constitutionality of Proposition 89 as applied to prisoners who were convicted of crimes committed before November 2, 1988. In Johnson v. Gomez, 92 F.3d 964, 965 (9th Cir.1996), California state prisoner Robert Johnson was convicted of first-degree murder in 1977 and given an indeterminate sentence of 25 years to life. After California voters passed Proposition 89, the Board found Johnson eligible for parole. Id. Under Proposition 89, Johnson was not eligible for release until the 30-day period for gubernatorial review had passed. Id. The Governor exercised his authority under Proposition 89 and reversed the grant of parole. Id. Johnson sought state habeas and his petition was denied. Id. Johnson then filed a federal habeas petition, which the district court denied. Id. at 966.9

We affirmed. Although we acknowledged that "the purpose and effect of the law here is to lengthen prison terms by making it more difficult for convicted murderers with indeterminate sentences to be released on parole," we held that this did not violate the Ex Post Facto Clause. Id. at 967. We ruled that "[t]he law . . . simply removes final parole decisionmaking authority from the [Board] and places it in the hands of the governor," who "must use the same criteria" as the Board. Id. We also noted that Johnson had not shown that he would have received parole under the old system and, therefore, Proposition 89 presented only a speculative risk of unconstitutionally lengthening his period of incarceration. Id. at 967-68.

Johnson controls here. The district court found that the evidence presented at trial showed that Proposition 89, "in actual practice, is not [a] `neutral' transfer of final decision-making authority from one decision-maker to another. . . . [W]hile the governors could use the law to review parole decisions to ensure that they are accurate and fair, they appear to have no such concern about decisions that deny parole." However, as we noted in Johnson, the Governor must use the same criteria to determine suitability as does the Board. Id. If the district court's finding that "governors have used [Proposition 89] to tip the scales against parole," is a finding that California Governors are not obeying state law, that finding is clearly erroneous. "Absent a demonstration to the contrary, we presume [state actors] follow[] . . . statutory commands." Garner, 529 U.S. at 256, 120 S.Ct. 1362. The district court did not point to evidence that Governors had reversed the Board other than "on the basis of the same factors which the parole authority is required to consider." Cal. Const. art. V, § 8(b). Nor did Gilman offer evidence showing that he would have received parole before the enactment of Proposition 89, and that Proposition 89 changed that result. Therefore, Proposition 89 remains only a transfer of decisionmaking power, which does not violate the Ex Post Facto Clause.

The district court erred in finding that Johnson does not control the outcome of Gilman's challenge to Proposition 89. We reverse the district court's findings and injunction as to Proposition 89, as to which Gilman is not entitled to relief.10

B. Proposition 9

We turn now to Proposition 9. The parameters for the district court's inquiry were set out in Gilman I. In that case, we reversed the district court's preliminary injunction against the enforcement of Proposition 9 because, although Proposition 9 "appear[ed] to `create[] a significant risk of prolonging [Plaintiffs'] incarceration,'" Gilman I, 638 F.3d at 1108 (second and third alterations in original) (quoting Garner, 529 U.S. at 251, 120 S.Ct. 1362), the PTA process allowed an inmate to advance his parole hearing if he could demonstrate a "change in circumstances or new information that establishes a reasonable likelihood that consideration of the public safety does not require the additional period of incarceration," id. at 1109 (quoting Cal.Penal Code § 3041.5(d)(1)).

At the outset, we note that proving a significant risk of prolonged incarceration in parole cases requires exacting evidence. The district court spent many pages of its opinion establishing, based principally on statistics derived from the Rutherford litigation, that Proposition 9 likely reduced the frequency of parole hearings for class members. That result is not surprising. However, a decrease in the frequency of parole hearings—without more—is not sufficient to prove a significant risk of lengthened incarceration. Id. at 1106. Proof of that risk is not a speculative inquiry. In California Department of Corrections v. Morales, 514 U.S. 499, 508-09, 115 S.Ct. 1597, 131 L.Ed.2d 588 (1995), the Supreme Court rejected the argument that a rule change violates the Ex Post Facto Clause if it "might create some speculative, attenuated risk of affecting a prisoner's actual term of confinement by making it more difficult for him to make a persuasive case for early release." "[C]onjectural effects are insufficient under any threshold we might establish under the Ex Post Facto Clause." Id. at 509, 115 S.Ct. 1597.

Moreover, Gilman must prove that Proposition 9 creates a significant risk of lengthened incarceration "within the whole context of [California's] parole system," Garner, 529 U.S. at 252, 120 S.Ct. 1362, including the opportunity for relief offered by the PTA process. In Gilman I, Gilman had urged four reasons why the PTA process was inadequate on its face to remove a significant risk of prolonged incarceration. 638 F.3d at 1109-10. We dealt with each one and found that none was established facially. Id. On remand, to prevail on his Ex Post Facto Clause claim, Gilman's task was to prove, by evidence drawn from Cal.Penal Code § 3041.5(d)(1)'s practical implementation, that the rule, as applied to him and other class members, did "not sufficiently reduce the risk of increased punishment for prisoners." Id. at 1109-11; see Garner, 529 U.S. at 255, 120 S.Ct. 1362.

To accomplish this task, Gilman marshaled evidence of grants and denials of PTAs. This evidence included cases in which (1) the PTA was granted and, at the consequent advance hearing, parole was granted; (2) the PTA was granted, but parole was ultimately denied; and (3) the PTA was denied, resulting in no advance hearing. Based on this evidence, the district court concluded that "[t]he PTA process is structured such that it fails, in many cases, to afford inmates a fair opportunity to obtain an advance hearing," and it "is not sufficient to protect inmates from the ex post facto problems inherent in Proposition 9."11

To reach this conclusion, the district court first reviewed the PTA process and decided for itself that "the advance hearing process sometimes works and sometimes does not work," because it "appears to deny advance hearings . . . to those who facially appear to deserve them." It then found that certain structural features of the PTA process created impediments to its proper functioning, rendering the PTA process illusory for some class members. However, the district court based these findings largely on speculation and inference from anecdotal evidence, rather than evidence drawn from Cal.Penal Code § 3041.5(d)(1)'s practical implementation proving that the PTA process failed to alleviate the classwide risk of lengthened incarceration posed by Proposition 9. See Garner, 529 U.S. at 255, 120 S.Ct. 1362; Morales, 514 U.S. at 508-09, 115 S.Ct. 1597; Gilman I, 638 F.3d at 1106. Because the district court applied the wrong standard, it committed legal error, and the resulting factual findings are clearly erroneous. We take each of the district court's points in turn to demonstrate the errors in its analysis.

1. Improper Denial of PTAs

In reviewing the PTA process, the district court first satisfied itself that "the PTA system works at denying petitions that ought to be denied," and then turned to the question "whether it grants petitions that ought to be granted." That the district court should so see its task is curious in light of its (correct) understanding that it "does not sit to review individual parole decisions." Notwithstanding this recognition of its limited review role, the district court went on to consider whether petitions were denied that "ought to [have been] granted."

After reviewing certain case studies, the district court concluded: "[T]he PTA [process] appears to deny advance hearings . . . to those who facially appear to deserve them." But, to decide who "facially" deserves the grant of a PTA, one must consider the merits of the grounds upon which the PTA is made. The question whether those grounds merit the grant of a PTA—like the question whether to grant parole—is committed to the sole "unfettered discretion" of the Board. In re Vicks, 56 Cal.4th 274, 153 Cal.Rptr.3d 471, 295 P.3d 863, 882 (2013); see Cal.Penal Code § 3041.5(d)(2); see also Swarthout v. Cooke, 562 U.S. 216, 220, 131 S.Ct. 859, 178 L.Ed.2d 732 (2011) (per curiam) (discussing the limits of federal review of a state's discretionary decisions with respect to the grant or denial of parole). Indeed, as Garner teaches us, in reviewing decisions of state parole authorities for potential Ex Post Facto Clause issues, the question is not whether "discretion has been changed in its exercise" by changes in parole procedures, but whether discretion "will not be exercised at all." 529 U.S. at 254, 120 S.Ct. 1362.12 It is undisputed that the Board did exercise its discretion as to each of the PTAs in question, granting some and denying others. We conclude that the district court erred in using its disagreement with the Board's decisions about which PTAs ought to have been granted or denied as a valid basis for finding an Ex Post Facto Clause violation.13

2. "Amorphous" Burden on Inmates

The district court then identified and examined various "structural problems" that could account for the Board's denial of PTAs. The most troubling problem found by the district court was that Cal.Penal Code § 3041.5(d)(1) places a new, additional, and "amorphous" burden on prisoners seeking an advance hearing—to show a "`change in circumstances or new information'. . . before [the Board] will even consider the question of suitability for parole." The district court found that the Board has interpreted the statute to require a "change in circumstances or new information" "in away that separates the `change in circumstances or new information' from the question of suitability." (Emphasis added.) This is an unnatural reading of Cal.Penal Code § 3041.5(d)(1), which specifically ties the "change in circumstances or new information" to that which "establishes a reasonable likelihood that consideration of the public safety does not require the additional period of incarceration of the inmate"; that is, suitability for parole.

Moreover, the evidence in the record favors this commonsense reading of the statute and runs counter to the district court's finding. A non-exclusive list of examples included in the Board's training materials suggests that the Board interprets Cal.Penal Code § 3041.5(d)(1) broadly. To show new information or a changed circumstance, an inmate need present only one or more of the building blocks that could result in a suitability finding, such as an updated parole plan, a job offer, completion of a substance-abuse-treatment program, or attainment of an educational certificate. The Board also requires, as does the statute, that the new information or changed circumstance "establishes a reasonable likelihood that consideration of public safety does not require an additional period of incarceration"—that is, that the new information or changed circumstance is reasonably likely to result in a finding of suitability for parole. In other words, if the Board follows its manual, it will deny an advance hearing only if it concludes that the inmate is unlikely to be found suitable for parole in light of all the information presented.14 Such an inmate—one who is likely unsuitable for parole—by definition is likely not to have received parole before the enactment of Proposition 9. And an inmate who was unsuitable a year ago and as to whom nothing has changed, similarly, was as unlikely to obtain parole before Proposition 9 as he is after. Cf. Morales, 514 U.S. at 512, 115 S.Ct. 1597 ("For these prisoners, the amendment simply allows the Board to avoid the futility of going through the motions of reannouncing its denial of parole suitability on a yearly basis.").

The district court concluded that the Board separated the "change in circumstances or new information" from parole suitability based on "some examples" of PTA denials, in particular, the denial of prisoner M. Brodheim's PTA.15 Brodheim did not procure an advance hearing through the PTA process. He obtained a hearing by successfully petitioning the district court for a writ of habeas corpus; we ultimately reversed the decision to grant the writ. While the appeal of that decision was pending, the Board found Brodheim suitable for parole but, after we reversed the improperly granted writ, Brodheim was returned to the same status as had obtained before the district court's action. He then filed a PTA, appending the transcript from the parole hearing at which he was found suitable for parole, but his PTA was summarily denied.

It may seem an abuse of discretion to have returned Brodheim to non-parole status after a hearing had determined him suitable for parole, even though that hearing was undeserved. Indeed, Brodheim may have had a state-law remedy, as Proposition 9 allows PTA denials to be "review[ed] by a court or magistrate . . . for a manifest abuse of discretion by the board." Cal.Penal Code § 3041.5(d)(2). However, we do not think it is possible to extrapolate from this single, peculiar example a finding as to how the Board handled PTAs from Gilman and other class members. The district court's inference, based solely on the Brodheim example, that after Proposition 9, class members face "incarceration indefinitely, unless the Board finds clear and convincing evidence of (a) a change in circumstances or new information, and separately, (b) suitability," is thus erroneous.

3. Decisions Passed to Next Panel

The district court found that there was a structural barrier to relief through the PTA process because decisionmakers denied a few PTAs without explaining whether there was a reasonable likelihood that further incarceration was not needed. According to the district court, this "tend[s] to show that the [Board] viewed certain issues as categorically exempt from the PTA process, and therefore could only be decided by panels after the deferral period imposed by the last panel."

Here, the district court assumed the Board was required to determine whether an inmate was suitable for parole whenever he filed a PTA because "that was the only question [the decisionmaker] had to decide." The district court did not consider whether the PTAs it referenced satisfied the statutory prerequisite: a "change in circumstances or new information" regarding suitability. Cal.Penal Code § 3041.5(d)(1). Indeed, it summarily dismissed the Board's finding that the statutory prerequisite had not been met by labeling the language the decisionmaker checked on the form order denying the PTA as "boilerplate":

That the Board may have decided to give additional reasons for its decision in other portions of its form order does not mean that the checked box's language is inadequate to establish that the Board made a decision on suitability, there and then. It does not mean that the Board made no decision as to the PTA by engaging in a "categorical exemption," as found by the district court in its claim that the Board had not exercised its discretion.16

In the example of a "boilerplate" denial cited by the district court, inmate T. Nguyen's "reason for denial" was first that he had not met the "change in circumstance or new information" requirement regarding suitability, and then that, not having met the requirement, the next regularly scheduled panel could consider his parole-suitability factors. The district court's reading of Nguyen's PTA denial was selective and does not support the inference that requiring further findings at the next scheduled hearing is evidence that there is a "categorical exemption" structural barrier to PTA grants, whereby the Board does not exercise its discretion to deny a PTA, but simply passes the decision to the next regularly scheduled hearing.17

4. Lack of Annual Comprehensive Risk Assessments

As to some inmates whose mental condition is a factor as to their suitability for parole, a psychological report, the CRA, is prepared only once every five years by State-employed personnel. Since the CRA is prepared only every five years, the district court found that a PTA would not be granted where the inmate had a psychological component to be evaluated to determine his suitability. It quoted Gilman's summation that "any prisoner who is denied parole in part because of the CRA has no chance of obtaining an advanced hearing." This finding was in error for two reasons. First, nothing prohibits a prisoner from procuring his own CRA using private resources. Second, the State points out that a CRA is not required for a PTA; the inmate-petitioner can address whatever issues were in his previous CRA through a personal statement to the Board, self-help programming, or evidence other than a psychological report.

5. Lack of Translation Services

The district court found a structural defect in the PTA process based on one inmate's claim that, after his petition with some Spanish language documents had passed an initial review, full review was denied until the documents were translated. However, the district court noted evidence that prisoners who need translation services are given such assistance with their PTAs. Rather than making a finding that class members were denied translation services, the district court found no facts to sustain its determination that translation services are unavailable. Instead, the district court ruled: "If in fact, no translation services are provided at the PTA stage, then the PTA process is illusory for those prisoners who communicate only in Spanish." (Emphasis added.) Quite obviously, unless a fact is found to exist, the supposition that it might exist is not a basis for decision.

6. Conclusions as to the PTA Process

The district court committed legal error by basing its findings principally on speculation and inference, rather than concrete evidence demonstrating that the PTA process failed to afford relief from the classwide risk of lengthened incarceration posed by Proposition 9. It erred by substituting its own judgment for the Board's regarding which PTAs ought to be granted. And the district court's findings of "structural problems" in the PTA process lack sufficient support in the record. The remaining findings, viewed under the correct legal standard, are insufficient to support a conclusion that, on this record, an as-applied Ex Post Facto Clause violation has occurred. We therefore reverse the district court's findings and injunction as to Proposition 9.

IV. Conclusion

For these reasons, we reverse the judgment of the district court and order the district court to enter judgment for the State of California.

REVERSED and REMANDED with instructions.