ORS 144.125C3)1 provides that the Board of Parole and Post-Prison Supervision (board) may postpone a prisoner’s scheduled release date if the board finds that the prisoner has a “present severe emotional disturbance such as to constitute a danger to the health or safety of the community.” In this case, the board found by a preponderance of the evidence that petitioner had such a condition, and petitioner sought judicial review arguing that the Due Process Clause required the board to apply a higher standard of proof. The Court of Appeals affirmed the board’s order without opinion. Stogsdill v. Board of Parole, 204 Or App 779, 132 P3d 62 (2006). We allowed review and now affirm the Court of Appeals decision.
In 1986, petitioner shot and killed his estranged wife’s boyfriend, struck his wife with the gun, and then shot another woman. As a result of those acts, petitioner pled guilty to murder, second-degree assault, and fourth-degree assault. The trial court sentenced petitioner to life imprisonment with a 10-year mandatory minimum sentence on the murder conviction and two lesser, concurrent sentences on the assault convictions.
Pursuant to ORS 144.120, the board set an initial release date for petitioner of February 5, 2000.2 In 2000, the board postponed petitioner’s release date, pursuant to ORS 144.125(3),3 until December 5,2002. In 2002, the board again *335considered whether to postpone petitioner’s release date. During an “exit interview,” the board asked petitioner about the crimes that he had committed and his progress within the institution. In addition to petitioner’s answers, the board had before it a recent psychological evaluation of petitioner, evidence from earlier hearings, and testimony from the sister of the person whom petitioner had killed. Based on that record, the board found by a preponderance of the evidence that petitioner “suffers from a present severe emotional disturbance that constitutes a danger to the health or safety of the community.” Pursuant to ORS 144.125(3), the board postponed petitioner’s release date for two years.
After exhausting his administrative remedies, petitioner filed a petition for judicial review of the board’s order postponing his release. Among other things, petitioner argued that the Due Process Clause required the board to find the requisite facts by clear and convincing evidence rather than by a preponderance of the evidence. As noted, the Court of Appeals affirmed the board’s order without opinion, and we allowed review to consider whether due process required the board to find the facts by clear and convincing evidence.4
Before turning to that federal constitutional issue, we begin with the subsidiary state law question of who bears the risk of nonpersuasion. See Davis v. Board of Parole, 341 Or 442, 445-47, 144 P3d 931 (2006) (analyzing subsidiary state law issues before deciding whether due process required proof by clear and convincing evidence). Two statutes bear on that analysis. ORS 144.120 directs the board to set an initial release date for most prisoners admitted to the Department of Corrections. See Engweiler v. Board of Parole, 340 Or 361, 367-71, 133 P3d 910 (2006) (discussing initial release dates). Another statute provides that, having set a release date, the board may postpone a scheduled release *336date if it finds that the prisoner has a “present severe emotional disturbance such as to constitute a danger to the health or safety of the community.” ORS 144.125(3).5
Under those statutes, petitioner is entitled to be released unless the board is persuaded that he has a present severe emotional disturbance that constitutes a danger to the health or safety of the community. See ORS 144.245 (explaining effect of a release date). Under that statutory scheme, petitioner has an interest in having the board employ a higher standard of proof. Compare Davis, 341 Or at 447-48 (holding that, under a different statutory scheme, the prisoner had no interest in having the board apply a higher standard of proof). That is, if due process requires proof by clear and convincing evidence, as petitioner argues, then petitioner would be entitled to be released if the board were persuaded by a preponderance of the evidence (but not by clear and convincing evidence) that he had a present severe emotional disturbance that made him a danger to the community. We accordingly turn to the question whether due process required the board to find the facts by clear and convincing evidence.
The Due Process Clause of the Fourteenth Amendment provides that no state shall “deprive any person of life, liberty, or property, without due process of law.” Procedural due process claims present two issues. See Wilkinson v. Austin, 545 US 209, 224, 125 S Ct 2384, 162 L Ed 2d 174 (2005) (explaining analysis). The first issue is whether the state has deprived a person of a liberty or property interest within the meaning of the Due Process Clause. Id. If it has, the second issue is what process is due. Id.
In this case, petitioner argues that the state statutes directing the board to set a release date created a protected liberty interest, which required the board to provide him with some process. As to that issue, we agree with petitioner. The United States Supreme Court has long recognized that state law can give rise to a constitutionally protected liberty interest. See Wolff v. McDonnell, 418 US 539, 557, 94 S Ct 2963, 41 *337L Ed 2d 935 (1974) (recognizing proposition). In 1995, the Court limited the types of state laws that would create protected liberty interests and overruled earlier decisions holding that mandatory state laws always created protected liberty interests. See Sandin v. Conner, 515 US 472, 483-84, 115 S Ct 2293, 132 L Ed 2d 418 (1995) (so holding). However, in overruling those decisions, the Court reaffirmed its decision in Wolff that a state law granting prisoners good time credits (which reduce the length of a prisoner’s sentence) created a constitutionally protected liberty interest, which the state could not revoke without providing some process. See id. at 480, 483 (reaffirming Wolff).
In this case, the board postponed petitioner’s release date. The board’s order resulted in the denial of petitioner’s early release from prison, much in the same way that the revocation of good time credits resulted in the denial of early release in Wolff. As the Court recognized in Wolff and reaffirmed in Sandin, a prisoner’s state-created interest in being released early from prison is an interest of “real substance,” which the state may not deny without due process. See id. The question that remains, however, is how much process is due. On that question, petitioner focuses solely on the standard of proof that the board employed.6 He argues that due process required the board to find the facts by clear and convincing evidence rather than by a preponderance of the evidence. We turn to that question.
The United States Supreme Court has employed a three-factor balancing test, first set out in Mathews v. Eldridge, 424 US 319, 96 S Ct 893, 47 L Ed 2d 18 (1976), to determine the standard of proof that due process requires. See Santosky v. Kramer, 455 US 745, 758, 102 S Ct 1388, 71 L Ed 2d 599 (1982) (applying that test to determine standard of proof required in parental termination proceedings). The Court explained that test in Mathews:
“□(Identification of the specific dictates of due process generally requires consideration of three distinct factors: First, the private interest that will be affected by the official *338action; second, the risk of an erroneous deprivation of such interest through the procedures used, and the probable value, if any, of additional or substitute procedural safeguards; and finally, the Government’s interest, including the function involved and the fiscal and administrative burdens that the additional or substitute procedural requirements would entail.”
424 US at 335.
Regarding the first factor, the private interest at stake, petitioner has a state-created right to be released on parole unless the board finds that he suffers from a present severe emotional disturbance that presents a danger to the health or safety of the community. Postponing a scheduled release date results in a substantial deprivation of liberty. In evaluating the significance of that deprivation, we note that the Court has distinguished among three related but separate deprivations: a criminal conviction, the revocation of parole, and the deprivation of good time credits. The Court has recognized that, of those three deprivations, the deprivation of good time credits is the least significant.
Regarding the revocation of parole, the Court has explained that parole arises after, and as a consequence of, a criminal conviction. Morrisey v. Brewer, 408 US 471, 480, 92 S Ct 2593, 33 L Ed 2d 484 (1972). It follows that revocation of parole “deprives an individual, not of the absolute liberty to which every citizen is entitled, but only of the conditional liberty properly dependent on observance of special parole restrictions.” Id. The revocation of parole, while a “grievous loss,” is of less weight in the Mathews balance than the loss of “absolute liberty” that results from a criminal conviction. See id. at 481.
Regarding the deprivation of good time credits, the Court has explained that, unlike the revocation of parole, “[t]he deprivation [of good time credits and the resulting denial of early release], very likely, does not then and there work any change in the conditions of [the prisoner’s] liberty.” Wolff, 418 US at 561. Rather, revocation of good time credits “postpone^] the date of eligibility for parole and extend [s] the maximum time to be served” unless the prison officials decide *339to restore the revoked credits. Id. Although the Court cautioned against “unrealistically discounting]” the significance of revoking good time credits, it concluded that revoking good time credits is “qualitatively and quantitatively different from the revocation of parole or probation.” Id.
The private interest at stake in this case falls somewhere between that implicated by revocation of parole and deprivation of good time credits. Although revoking good time credits may not always result in denying a prisoner early release, see Wolff, 418 US at 561 (recognizing that proposition), postponing a release date almost always will affect a prisoner’s actual date of parole. However, postponing the release date is similar to revoking good time credits in that it continues the prisoner’s existing status; it does not deprive a prisoner of either absolute or conditional freedom as do conviction and revocation of parole. It follows, we think, that the interest affected by postponing a release date is closer to the deprivation of good time credits than it is to the revocation of parole.
In evaluating the significance of petitioner’s interest, we also note that the stigma associated with a criminal conviction is far greater than the stigma associated with a determination that a prisoner is ineligible for early release. See In re Winship, 397 US 358, 373-74, 90 S Ct 1068, 25 L Ed 2d 368 (1970) (Harlan, J., concurring) (discussing stigma associated with delinquency determination in concluding that due process required heightened standard of proof). The board’s determination that continued confinement is appropriate because the prisoner has a condition that makes him or her a danger to the community simply reaffirms, in many respects, the original determination implicit in the judgment of conviction that confinement was necessary both for the prisoner’s reformation and the community’s safety.
Finally, we note that the board’s determination does not result in a permanent deprivation, as did the actions at issue in Santosky and Cruzan v. Director, Missouri Department of Health, 497 US 261, 110 S Ct 2841, 111 L Ed 2d 224 (1990). This is not a case in which the state is seeking to terminate parental rights, as in Santosky, nor is it a case in which clear and convincing proof that an incompetent person *340wanted to have life-sustaining medical treatment withdrawn was required, as in Cruzan. In both those instances, the Court recognized that an erroneous determination would not be “susceptible of correction,” and it focused on that fact in holding that a clear-and-convincing-evidence standard complied with due process. Cruzan, 497 US at 283; Santosky, 455 US at 759. In this case, by contrast, an erroneous decision does not result in a permanent deprivation. Rather, the board deferred petitioner’s release for a two-year period, and ORS 144.122 permitted petitioner to ask the board to “reset [the scheduled release date] to an earlier date” than that.
The second Mathews factor is “the risk of an erroneous deprivation of [the private] interest through the procedures used, and the probable value, if any, of additional or substitute procedural safeguards.” Mathews, 424 US at 335. In applying that factor to issue of the standard of proof, the Court has explained that “the minimum standard of proof tolerated by the due process requirement reflects not only the weight of the private and public interests affected, but also a societal judgment about how the risk of error should be distributed between the litigants.” Santosky, 455 US at 755. That is, in applying the second Mathews factor, the Court has not asked simply whether a higher standard of proof will reduce the risk of an erroneous deprivation (it always will); rather, the Court has looked to a “societal judgment” about how the risk of error should be distributed between the parties.
In making that determination, the Court has relied on the reasoning set out in Justice Harlan’s concurrence in Winship. See, e.g., Santosky, 455 US at 754-55. We follow the same course. Justice Harlan reasoned in his concurring opinion:
“[T]he trier of fact will sometimes, despite his best efforts, be wrong in his factual conclusions. In a lawsuit between two parties, a factual error can make a difference in one of two ways. First, it can result in a judgment in favor of the plaintiff when the true facts warrant a judgment for the defendant. The analogue in a criminal case would be the conviction of an innocent man. On the other hand, an erroneous factual determination can result in a judgment for the defendant when the true facts justify a judgment in *341plaintiffs favor. The criminal analogue would be the acquittal of a guilty man.
“The standard of proof influences the relative frequency of these two types of erroneous outcomes. If, for example, the standard of proof for a criminal trial were a preponderance of the evidence rather than proof beyond a reasonable doubt, there would be a smaller risk of factual errors that result in freeing guilty persons, but a far greater risk of factual errors that result in convicting the innocent. Because the standard of proof affects the comparative frequency of these two types of erroneous outcomes, the choice of the standard to be applied in a particular kind of litigation should, in a rational world, reflect an assessment of the comparative social disutility of each.”
Winship, 397 US at 370-71 (Harlan, J., concurring).
Justice Harlan explained that the standard of proof in a criminal prosecution — beyond a reasonable doubt— reflects the considered societal judgment that “it is far worse to convict an innocent man than to let a guilty man go free.” Id. at 372 (Harlan, J., concurring). Conversely, the preponderance-of-evidence standard in civil matters reflects a long-held understanding that it is “no more serious in general for there to be an erroneous verdict in the defendant’s favor than for there to be an erroneous verdict in the plaintiffs favor.” Id. at 371 (Harlan, J., concurring).
Here, the board applied a preponderance-of-the-evidence standard; it determined that the risk of error should be allocated equally between the board and the prisoner. Petitioner has not identified any basis for saying that there is a shared societal judgment that parole boards should employ a higher standard of proof and thus err on the side of releasing dangerous offenders early. He does not identify any other court that has held, in this context, that due process requires clear and convincing evidence, nor has he identified any other state that requires, as a matter of state law, proof by clear and convincing evidence that an inmate is not fit for early release. Compare Cooper v. Oklahoma, 517 US 348, 360, 116 S Ct 1373, 134 L Ed 2d 498 (1996) (looking to practices in other states in deciding what standard of proof due process requires); Santosky, 455 US at 749 (same).
*342The reason for that allocation derives, we think, from the existence of a conviction — a proposition that petitioner’s case illustrates. Petitioner has been convicted of murder and sentenced to life imprisonment. Consistently with due process, the state could confine him for the duration of his sentence, i.e., the remainder of his life. See Greenholtz v. Nebraska Penal Inmates, 442 US 1, 7-8, 99 S Ct 2100, 60 L Ed 2d 668 (1979) (recognizing that proposition). If the state may confine petitioner for the duration of his sentence, then it is difficult to see why the state may not allocate the risk of error equally in deciding whether he qualifies for early release. As the Court has recognized, due process does not require states either to provide for parole or to structure their parole systems in a particular way. See id. at 7-8 (recognizing that proposition).
The third Mathews factor is “the Government’s interest, including the function involved and the fiscal and administrative burdens that the additional or substitute procedural requirements would entail.” 424 US at 335. In evaluating the government’s interest, courts typically have focused on the burdens that providing additional procedural protections would entail. For example, in the prison context, the Court has recognized that, although the right to call witnesses is a basic tenet of due process, “[p]rison officials must have the necessary discretion to keep the hearing within reasonable bounds and to refuse to call witnesses that may create a risk of reprisal or undermine authority, as well as to limit access to other inmates to collect statements or compile other documentary evidence.” Wolff, 418 US at 566. In this case, however, the board has not identified comparable administrative considerations that a higher standard of proof would impose on either the board or prison officials, nor are we aware of any.
Considering the three Mathews factors, we note that the significance of petitioner’s private interest is diminished by the existence of a conviction that permits the state to hold him for the duration of his sentence and that he has not identified any basis for saying that the board should err on the side of releasing dangerous offenders early. Accordingly, we hold that due process does not require the board to apply a *343clear-and-convincing-evidence standard of proof. Proof by a preponderance of the evidence is sufficient.
The decision of the Court of Appeals and the order of the Board of Parole and Post-Prison Supervision are affirmed.