State, Department of Corrections v. Cowles

151 P.3d 353 (2006)

STATE of Alaska, DEPARTMENT OF CORRECTIONS, Petitioner,
v.
Ebony COWLES, Respondent.

No. S-11352.

Supreme Court of Alaska.

December 15, 2006.

Venable Vermont, Jr., Assistant Attorney General, Anchorage, and Gregg D. Renkes, Attorney General, Juneau, for Petitioner.

Charles W. Ray, Jr., Law Offices of Charles W. Ray, Jr., P.C., and Paul W. Whelan and Kevin Coluccio, Stritmatter Kessler Whelan Withey Coluccio, Seattle, Washington, for Respondent.

Before: BRYNER, Chief Justice, MATTHEWS, EASTAUGH, FABE, and CARPENETI, Justices.

OPINION

FABE, Justice.

I. INTRODUCTION

A parolee murdered his former girlfriend and then shot himself. The issue before us is whether the State of Alaska may be held liable in tort for a crime committed by a parolee under its supervision. The State urges us to overrule our holding in Division of Corrections v. Neakok that the State owes a duty of care in supervising its parolees.^[1] The State also contends that it is immune from suit and that it is entitled to judgment as a matter of law on the issue of causation. Because we find that at least some of the State's alleged acts of negligence are shielded by discretionary function immunity, we vacate the superior court's order denying the State's motion for summary judgment. We therefore overrule our holding in Neakok that the "selection of conditions" of parole are operational activities not entitled to immunity.^[2] While we decline to overrule our holding in Neakok that the State owes a duty of care to the victims of parolees under its supervision, we emphasize that this duty should be narrowly construed. Based on the facts of this case, we conclude that the trial court properly denied summary judgment on the issue of duty. We remand for a ruling on causation in light of the superior court's discretionary function immunity rulings.

II. FACTS AND PROCEEDINGS

In September 1991 Calvin McGrew and four others were driving a jeep that caught fire near the home of Jacqueline and Donald Boschert on the Parks Highway.^[3] The Boscherts helped them put out the fire and allowed the five people into their house to use the phone. McGrew grabbed Mrs. Boschert and put a knife to her throat. The group tied the Boscherts up, stole some personal items, and escaped in the Boscherts' Cadillac and Ford pickup truck.

McGrew pleaded no contest to robbery in the first degree and other charges. He received a presumptive seven-year sentence for the robbery charge.^[4] McGrew was released on mandatory parole on November 23, 1996 subject to thirteen general parole conditions and fourteen supplemental conditions.

When McGrew was released from prison, Patricia Beckner was assigned to serve as his parole officer. As required by Department of Corrections (DOC) policy, Beckner filled out a risk assessment form to determine McGrew's supervision level.^[5] Beckner failed to include McGrew's prior juvenile convictions in the calculation of the risk score. She also incorrectly assigned McGrew a "medium" supervision level even though the risk score total on the form corresponded to a "maximum" supervision level. Under DOC policy, medium supervision requires parolees to have face-to-face contact with the parole officer at least once a month, while maximum supervision requires twice monthly face-to-face contact and a field visit at least once every four months.^[6]

DOC policy also required that a risk reassessment be completed every six months.^[7] Like her initial risk assessment, Beckner's reassessments were incorrectly scored. Beckner also filled out a risk reassessment in August 1998 even though McGrew had absconded from parole at that time and she had not had contact with him for several months. At the time of the murder, Beckner had not completed a reassessment form for McGrew in more than ten months.

Between his release from prison in November 1996 and March 1998, McGrew appears to have complied with his parole conditions with the exception of a few missed appointments, positive urine tests for marijuana, and "no shows" for urinalysis appointments. During this time McGrew lived with his girlfriend Shila Davis or her parents.

McGrew did not appear for urine testing in March 1998 and failed to report to Beckner in April and June 1998. Beckner learned from McGrew's employer that he had been fired for not showing up. On July 17, 1998, Beckner filed a parole violation report. A parole board member issued a parole arrest warrant four days later. When McGrew had not been arrested by November 1998, Beckner put his file in "abscond" status.

McGrew was arrested on May 13, 1999. A parole board member conducted a preliminary hearing on May 21, 1999 to determine if he should remain in custody until his full board revocation hearing. At the hearing, McGrew admitted to five parole violations and presented a release plan proposing that he continue to work, live with Shila Davis, and support his family. Beckner submitted a letter from Davis asking that McGrew be released. The letter stated that she and McGrew had been together for seven years, that they had recently had twin daughters, and that she needed him at home to help raise their children. The parole board member released McGrew on May 21, 1999 with the additional condition that he present proof of employment to Beckner upon request.

On June 8, 1999, Davis filed a petition in the trial court requesting a protective order. She alleged that McGrew had abused her and requested that he be barred from contacting her or coming to her home. On June 13, 1999, Davis called 911 alleging McGrew had hit her. On July 23, 1999, Davis filed another petition for a protective order, and the district court granted an ex parte protective order. The next day, on July 24, 1999, McGrew shot Davis and himself. One or both of their bodies fell on their three-month-old twins. One of the twins survived, but the other suffocated.

At the time of the murder, parole officer Beckner was unaware that Davis had filed domestic violence petitions against McGrew or that the couple had separated. Beckner stated in her affidavit filed in the resulting tort case that while supervising McGrew between his release after the preliminary hearing in May and the murder in July, she "attached a great deal of significance to what appeared to be a long-term stable relationship, and the recent arrival of twins." Beckner further stated: "I felt that Calvin's relationship with Shila, and his continued employment, were good indicators that he might succeed on parole and afterwards." McGrew had reported to Beckner as scheduled on May 24 and June 21. He did not tell Beckner about the domestic violence problems. But on his June 21 monthly report form he did note that he had had contact with the police, stating "[t]hey came to my house to arrest me." In her affidavit, Beckner stated that she had "no current recollection of this entry or any discussion about it with Calvin, but [she] would have asked him about it and . . . feel[s] sure that [she] did." She stated that she believed that the visit from the police was related to "ongoing confusion" between McGrew and Darryl Poindexter^[8] and did not indicate to her that McGrew had committed new crimes or significant parole violations. In her deposition, Beckner stated that she remembered McGrew submitting this report and that she believed that the police visit was related to the "mix-up of identities" between McGrew and Darryl Pointdexter, but she also agreed with her earlier testimony that she had no specific recollection of McGrew's entry and that the police visit could have been in response to a domestic violence complaint by Shila Davis. In response to the June 21 report, Beckner took no action to find out from the police why they had tried to arrest him. On July 22, 1999, two days before the murder, McGrew failed to report as scheduled. Beckner took no action in response to the missed visit.

Ebony Cowles, the personal representative of the Estate of Shila Davis and guardian of the surviving child, filed suit against the State Department of Corrections and the Municipality of Anchorage. The complaint alleges that the State committed negligence by failing to implement and enforce an appropriate parole plan, to require appropriate post-release therapy, to enforce parole violations, to properly supervise McGrew, and to revoke his parole.

The State moved for summary judgment. The State argued that our decision in Neakok should be overruled and therefore that the State owes no duty of care to victims of crimes committed by parolees. The State also argued that it was immune from suit and that it was entitled to judgment as a matter of law on the issue of causation. The superior court denied the motion for summary judgment. We accepted the State's petition for review.

III. STANDARD OF REVIEW

"We review denials of summary judgment motions de novo to determine whether there are genuine issues of material fact and whether the moving party is entitled to judgment as a matter of law, viewing all facts in the light most favorable to the non-movant."^[9] We review de novo the existence and extent of a duty of care.^[10] Whether a governmental act is entitled to discretionary function immunity is also a matter of law that is reviewed de novo.^[11]

We overrule a prior decision of this court when "we are `clearly convinced that the rule was originally erroneous or is no longer sound because of changed conditions, and that more good than harm would result from a departure from precedent.'"^[12]

IV. DISCUSSION

We addressed in Neakok^[13] the question whether the State can be held liable for negligent parole supervision. In that case, the relatives of three people murdered by a mandatory parolee sued DOC for negligence, alleging, among other things, that DOC failed to impose appropriate parole conditions and to supervise the parolee adequately.^[14] We held in Neakok that the State owes a duty of care to a parolee's foreseeable victims and that the State is not immune from suit for negligence in the selection of parole conditions and the supervision of parolees.^[15] In this appeal the State asks us to overrule Neakok by finding that the State owes no duty of care to parolees' victims and that discretionary function immunity applies to all of Cowles's claims, including those alleging negligent imposition of parole conditions and negligent parole supervision. The State also argues that it was entitled to judgment as a matter of law on the issue of causation.

Although we usually determine whether a tort duty exists before reaching the question of discretionary function immunity,^[16] in this case we analyze immunity first because doing so clarifies the public policy considerations that also bear on our duty analysis.^[17]

A. Discretionary Function Immunity

Under the Alaska Tort Claims Act, the State is immune from suit for tort claims "based upon the exercise or performance or the failure to exercise or perform a discretionary function or duty on the part of a state agency or an employee of the state, whether or not the discretion involved is abused."^[18] We have recognized that "the term `discretionary' in AS 09.50.250 should not be interpreted broadly to encompass all state actions involving discretion."^[19] Instead, we examine each act or function to determine whether it can be described as "planning" or "operational."^[20] "A planning decision is one that involves policy formulation," while an operational decision involves the execution or implementation of a policy already formulated.^[21] When statutes and regulations do not require officials to execute a predetermined policy and instead delegate authority to decide policy matters, the resulting determinations are planning decisions.^[22] "Only acts or functions occurring at the planning level are entitled to immunity as discretionary functions under AS 09.50.250."^[23] In this way, discretionary function immunity "ensures that courts do not step into the policy roles committed to other branches of government."^[24]

The State argues that discretionary function immunity bars all claims in this suit. Cowles responds that the State is not immune because parole supervision involves operational acts rather than planning decisions. We do not determine whether an entire class of decisions, such as those related to parole supervision, are entitled to immunity.^[25] Rather, the allegedly negligent decisions in a particular case must be examined individually to determine if they are "planning" or "operational" in nature.^[26] Cowles's complaint alleges that the State and DOC failed to implement and enforce an appropriate parole plan for McGrew, to require appropriate post-release therapy, to enforce and report parole violations, to act in response to McGrew's dangerous behavior, to comply with the statutes, regulations, and guidelines governing supervision of parolees, and to revoke McGrew's parole. The parties have not systematically addressed each of these individual allegations in their briefs. We set forth here the boundaries of discretionary function immunity as it relates to the decisions involving parole supervision addressed in Cowles's complaint. On remand, the superior court should make a separate determination for each of Cowles's allegations following the principles announced in this opinion and in our other decisions concerning discretionary function immunity.^[27] The allegations of negligence in Cowles's complaint can be placed into two groups for the purpose of the discretionary duty analysis: decisions made by the parole board or its members, and decisions made by McGrew's parole officer.

1. Decisions made by the parole board

Some allegations in Cowles's complaint appear to refer to actions or omissions on the part of the parole board. These include failure to impose appropriate parole conditions, to require appropriate post-release therapy,^[28] and to revoke McGrew's parole.^[29] All of these decisions are protected by discretionary function immunity. As pointed out in the dissent in Neakok written by Justice Matthews and joined by Chief Justice Rabinowitz, decisions regarding conditions of parole "are akin to decisions which a sentencing judge must make in deciding on the terms of a sentence of probation and are plainly discretionary."^[30]

Determining the appropriate parole conditions upon release and deciding whether or not to revoke an offender's parole or whether or not to release an offender pending a revocation hearing are all planning decisions. They require careful weighing of policy considerations, including public safety, the need to rehabilitate and reintegrate offenders, the allocation of resources available to treat and supervise parolees, and potential prison overcrowding.^[31] The purpose of discretionary function immunity — to maintain the separation of powers between the judiciary and the executive or legislative branches and to give agencies latitude to perform their policy-making functions without fear of liability^[32] — dictates that such determinations be shielded with immunity. We therefore overrule the holding in Neakok that "[f]ormulation of [a] parole plan, and selection of special [parole] conditions" are not planning activities entitled to immunity.^[33]

In Neakok, our analysis of duty and immunity applied only to parole officers, not the parole board, because no reasonable jury could find the board at fault in that case.^[34] But the language in Neakok was broad. Because the parole board makes decisions about parole conditions, Neakok could be read to apply to the board. We therefore overrule this holding to ensure that officials are not pressured to "err on the side of restrictiveness when considering discretionary parole."^[35] We are clearly convinced that this rule was originally erroneous and that departing from our holding in Neakok would do more good than harm by preventing courts from intruding on policy-making activities committed to other branches of government.^[36]

2. Decisions made by the parole officer

Cowles's complaint also alleges that Beckner committed negligence by failing to enforce and report parole violations, to comply with the statutes, regulations, and guidelines governing the supervision of parolees, to act in response to McGrew's dangerous behavior, and to seek to revoke his parole. The parties dispute whether Beckner's acts were discretionary or operational. In asserting that the State is not entitled to immunity, Cowles focuses on three allegedly negligent omissions on the part of Beckner: (1) failure to report parole violations and seek revocation; (2) failure to follow DOC policies governing offender classification and supervision standards; and (3) failure to investigate potential parole violations and ensure McGrew's compliance with the conditions of parole.

Cowles argues that Beckner was required by DOC policy to respond to McGrew's parole violations, and therefore that her failure to recommend parole revocation was not immune. The State contends that a parole officer's decision to report a parole violation or to pursue revocation is immune from suit because it requires judgment and discretion to determine "the point at which counseling, . . . warnings and other means to obtain compliance with conditions are unsuccessful." The State's characterization of the parole officer's function is accurate with respect to the officer's response to minor parole violations, but not serious violations. DOC Policy 902.14 mandates that an officer "shall" file a petition to revoke parole if there is probable cause that the offender committed a "serious violation" but is given discretion whether to initiate revocation action if the offender commits a "minor violation."^[37] Under the policy, a "serious violation" includes all felony behavior, "[c]lass A misdemeanor(s), except in the instance where the supervising probation officer's discretionary authority may denominate the misdemeanor as a minor violation," and technical violations that constitute a criminal act or jeopardize the property or safety of another person.^[38] Minor violations include class B misdemeanors, state or municipal code violations, and technical parole violations.^[39]

A parole officer therefore has no discretion, and thus no immunity, in responding to felony behavior or other actions that jeopardize the property or the safety of another person; in such cases the officer is simply executing a pre-existing policy. However, the officer can exercise judgment in deciding whether to petition to revoke parole where a client commits a minor violation. The exercise of some discretion does not in itself confer immunity.^[40] Nevertheless, when the parole officer is given a choice, the decision whether or not to seek to revoke parole involves the same weighing of policy matters that a parole board engages in when it makes the final parole revocation decision. Beckner was not aware of a parole violation that required her to petition to revoke McGrew's parole under DOC policy.^[41] Her decision not to pursue revocation in response to McGrew's technical violations, including his missed appointment two days before the murder, is therefore protected by discretionary function immunity.

Cowles's complaint also alleges that Beckner failed to comply with the regulations and guidelines governing parole supervision. She argues on appeal that no immunity attaches to Beckner's failure to accurately complete the risk assessments for McGrew and to follow up on the statement on his monthly report form that the police had come to his house to arrest him. The day-to-day supervisory activities of a parole officer, such as filling out risk assessment scales and investigating the apparent commission of a serious violation are operational duties not entitled to immunity. DOC policies mandate that the officer use the risk assessment scales to assign a supervision level to each case, and set forth the extent and type of contact required to supervise offenders based on their risk classification.^[42] Similarly, DOC Policy 902.14 requires officers to investigate the charges and circumstances surrounding "[a]n offender's arrest for, or apparent commission of," a serious or minor violation, and to seek revocation if there is probable cause that the offender committed a serious violation.^[43] Discretionary function immunity does not apply to these activities because a parole officer is not required to choose between competing policy concerns in performing these duties, but merely to exercise some judgment in carrying out established DOC directives.^[44] This conclusion finds support in the decisions of other courts that have refused to apply discretionary function immunity to the day-to-day functions of probation and parole officers.^[45] Accordingly, the State is not immune from suit based on Beckner's alleged negligence in filling out the risk assessment forms. It is unclear whether Beckner was given notice of the apparent commission of a violation because there is a question of fact as to what happened at the meeting at which McGrew gave Beckner the monthly report stating that the police had come to arrest him and whether, based on that interaction, Beckner could conclude that the report did not indicate an apparent commission of a parole violation. Because this case reaches us from a denial of summary judgment, we must view all facts in the light most favorable to the non-moving party. Issues of material fact preclude summary judgment for the State on this issue.

However, seeking out possible parole violations of which the parole officer has no notice involves planning decisions that are entitled to discretionary function immunity. A parole officer must make policy judgments in deciding how to allocate time and resources among various clients.^[46] And the officer must balance the interests of public safety and rehabilitation of offenders when deciding how much time to devote to seeking out potential parole violations as opposed to assisting clients with housing, rehabilitation, and other needs.^[47] We follow Justice Matthews's Neakok dissent in declining to impose liability "for an alleged failure to seek out parole violations."^[48] Accordingly, the State may not be held liable for Beckner's alleged negligence in failing to take affirmative action to discover parole violations of which she had no notice.

B. Duty

The State contends that it owes no duty of care to the victims of crimes committed by parolees. It urges us to overrule our contrary holding in Neakok, arguing that the reasoning of our decision in Sandsness^[49] suggests that this part of Neakok is erroneous.

In Sandsness, a juvenile offender committed murder not long after he had been released from custody.^[50] The victim's relatives sued the State for negligence, alleging that the State knew the juvenile posed a threat and should have asked the court to extend his commitment.^[51] We held that the State had no duty to use due care in deciding whether to extend the commitment of a juvenile offender.^[52] We reasoned that imposing such a duty would conflict with the State's goal of rehabilitating juvenile offenders by causing the State to "`err on the side of excessive detention to avoid negligence suits or judgments.'"^[53] The State argues that the same concerns animate this case. But Sandsness involved a claim of negligent release, not negligent parole supervision.^[54] Unlike the situation in Sandsness, there is no danger that the prospect of state tort liability will be a factor in deciding whether to incarcerate an offender. The State is immune from liability for negligence in the parole board's decisions concerning parole conditions and parole revocation. Allowing recovery for a parole officer's failure to comply with DOC policies governing parole supervision does no more than encourage adherence to those policies, reinforcing rather than distorting the balance between public safety and rehabilitation fashioned by the legislature and DOC.

The State points out that in Sandsness we relied on the Vermont Supreme Court in Sorge v. State^[55] and argues that we should adopt the reasoning in Sorge to overrule Neakok. The Sorge court declined to find a duty in a non-detention juvenile corrections context because the primary function of probation and parole is to "rehabilitate conduct rather than control it."^[56] The reasoning of Sorge was relevant to our holding in Sandsness because Sandsness was a negligent release case that presented the danger that imposing liability would result in excessive detention. But our approach in the present case is equally protective of the rehabilitative function of supervision programs as that used in Sorge. Sorge uses a blunt instrument to protect the rehabilitative goal of parole programs: it declines to impose liability in any context by refusing to impose a duty to supervise parolees. We adopt a more targeted approach by imposing such a duty but using discretionary function immunity to ensure that the rehabilitative purpose of parole will not be jeopardized.

Nor are we "clearly convinced"^[57] that our determination in Neakok that the State has a duty to exercise due care in supervising parolees is erroneous. The proposed final draft of the Restatement (Third) of Torts includes a comment explicitly stating that "those who supervise parolees, probationers, or others in prerelease programs . . . are appropriately held to an affirmative duty of reasonable care."^[58] Although there is a split of authority on this question, several courts in other jurisdictions have imposed a duty of care for the supervision of parolees.^[59]

At least one court has suggested that before liability can attach the parole officer must have a special relationship to the victim or victim class that creates a duty beyond that owed to the public as a whole.^[60] The "victim class" in that case was defined to include young boys who might have frequent contact with the offender.^[61] This approach stems from the importance of foreseeable harm in the determination of whether to impose a tort duty.^[62] In Neakok, we held that a duty of care is owed to victims who are foreseeable, though not necessarily specifically identifiable.^[63] We noted that in that case the victims "were foreseeable as more than simply members of the general public" because they were residents of the isolated community into which the offender was released, and because one of the victims had been identified by a prison employee as particularly at risk.^[64] Forgiving the opinion's broad dicta, we read Neakok to impose a duty only where officials know, or reasonably should know, that a parolee poses a danger to a particular individual or identifiable group. Thus, our foreseeability analysis in Neakok is quite similar to the requirement of a particularized relationship to an identifiable "victim class" adopted by some jurisdictions and leads to similar results. McGrew's victims were foreseeable and identifiable as more than simply members of the general public because they were members of McGrew's household. Similarly, in the companion case C.J. v. State, Department of Corrections, it was foreseeable that if not properly supervised, the offender, a convicted rapist, might rape a woman in the community into which he was released.^[65]

The dissent in Neakok suggests that a duty should be imposed only where a parole officer receives notice of imminent peril, akin to our cases involving the duty of a police officer to respond to a known life-threatening situation.^[66] But parole officers, unlike police officers, have a special relationship to parolees under their supervision that gives rise to a duty of care.^[67] The closest analogy to the Neakok dissent's approach is our decision in Dinsmore-Poff v. Alvord, where we held that parents who know that their child has a tendency towards violence can be held liable for harm committed by that child only if "the parent had reason to know with some specificity of a present opportunity and need to restrain the child to prevent some imminently foreseeable harm."^[68] But parents are in quite a different position from parole officers. Parole officers are trained professionals employed by an agency whose job it is to formulate supervision policies that promote rehabilitation while protecting the public from the offenders in its charge. Most important, unlike parents, the State is immune from liability for all discretionary planning or policy decisions made by parole officers; liability will attach only where a parole officer commits negligence in performing operational functions that implement DOC's policies.

The dissent suggests that liability for negligent supervision could lead the State to err on the side of continued detention to avoid liability.^[69] But discretionary decisions about parole release, conditions, and revocations are immune from liability. Because neither the parole board nor parole officers are exposed to liability for policy decisions to release, set conditions, or revoke parole, our decision creates no incentive for additional confinement. And if the State is concerned about possible liability from negligent supervision associated with officers' operational duties, adequate training can offer a mechanism to ensure that officers adhere to department policies.^[70] We recognize that parole supervision is a difficult job. But we believe that imposing a duty of care while providing for discretionary function immunity for policy decisions will protect the State's rehabilitative goals while encouraging parole officers to carry out their mandated operational duties in a non-negligent manner.

Nothing in our case law or the cases in other jurisdictions convinces us that our duty analysis in Neakok is erroneous. We therefore decline to overrule our holding in Neakok that the State has an actionable duty of care in supervising parolees. Our case law has established factors, commonly referred to as the D.S.W. factors, to determine whether a duty of care exists.^[71] But as the State has recognized, we adhere to precedent without re-applying the D.S.W. factors when we have already addressed the issue of duty in a closely related case.^[72]Neakok squarely addressed the issue of negligent supervision and established a duty to exercise due care in supervising parolees. Having found that this holding does not meet the standard for overturning precedent, we decline to engage in the dissent's analysis of the D.S.W. factors as if Neakok had not been decided. The State appealed the trial court's denial of summary judgment on all three asserted independent grounds: duty, discretionary function immunity, and causation. Even though we uphold the trial court's denial of summary judgment on the issue of duty, we are still compelled to examine its ruling on discretionary function immunity. And as the State argued in its summary judgment motion, "discretionary function immunity addresses the same public policy issues found in the duty analysis." Thus, we address many of the same public policy implications raised by a duty analysis through our discussion of discretionary function immunity.

We therefore affirm the trial court's ruling on the issue of duty: there is sufficient evidence to withstand summary judgment. When viewed in the light most favorable to Cowles, the record suggests that Beckner knew of the relationship between Davis and McGrew and knew that the police had come to arrest McGrew. This evidence raises a genuine issue of material fact as to whether Beckner knew or reasonably should have known that McGrew posed a danger to Davis and her family and whether Beckner breached her duty.

C. Causation

The State argues that even if it breached a duty to Cowles and is not immune from liability for that breach it is nevertheless entitled to summary judgment because as a matter of law its actions did not cause Cowles's injuries. Cowles argues that the issue of causation is a question of fact for the jury.

We have held that "negligent conduct may properly be found to be a `legal cause' of a plaintiff's injury if the negligent act was more likely than not a substantial factor in bringing about (the) injury."<sup>[73]</sup> The substantial factor test is satisfied by showing "`both that the [injury] would not have happened "but for" the defendant's negligence and that the negligent act was so important in bringing about the injury that reasonable men would regard it as a cause and attach responsibility to it.'"^[74]

The State contends that the connection between its acts and the murder was too attenuated to establish causation, particularly since Beckner could not have revoked McGrew's parole without the independent action of the parole board. Our decision requires the superior court to reexamine whether the State is entitled to discretionary function immunity for some of its allegedly negligent acts. That reexamination may affect the superior court's causation analysis, for Cowles may only rely on non-immune acts to establish causation. We therefore remand the case to the superior court for a ruling on causation in light of its discretionary function immunity rulings. Although Cowles's theory of liability requires an extended chain of causation, we cannot say that as a matter of law the State's alleged negligence did not cause her injuries. Reasonable jurors could find that Beckner would have discovered the domestic violence complaints if she had supervised McGrew properly. They could further find that this information would have changed Beckner's and the parole board's view of McGrew since, according to her affidavit, Beckner "attached a great deal of significance" to what she thought was a long-term stable relationship with Davis indicating he might succeed on parole. Reasonable jurors could find that once the domestic violence problem was known the parole board would likely have revoked McGrew's parole, thereby preventing the murder. Accordingly, material questions of fact preclude summary judgment for the State on the issue of causation.

V. CONCLUSION

Because the State is entitled to discretionary function immunity with respect to at least some of the allegations in the complaint, we VACATE the superior court's order denying the State's motion for summary judgment. We REMAND for further proceedings consistent with this opinion.

BRYNER, Chief Justice, with whom CARPENETI, Justice, joins except with regard to the last paragraph, concurring.

MATTHEWS, Justice, dissenting in part.

BRYNER, Chief Justice, with whom CARPENETI, Justice, joins except with regard to the last paragraph, concurring.

I agree with the court's decision but have reservations concerning three points addressed in its opinion.

First, I do not believe that it is necessary to overrule any aspect of Neakok^[1] in this case. Because any decision that overturns precedent necessarily erodes the values of stare decisis, I would use this extraordinary power only upon a compelling showing that its use is necessary and only when the point of law to be disavowed is squarely at issue.

As we have consistently recognized in the past, "a party raising a claim controlled by an existing decision bears a heavy threshold burden of showing compelling reasons for reconsidering the prior ruling: `We will overrule a prior decision only when clearly convinced that the rule was originally erroneous or is no longer sound because of changed conditions, and that more good than harm would result from a departure from precedent.'"<sup>[2]</sup> Here, the state has failed to meet this burden.<sup>[3]</sup> Although it insists that Neakok was wrongly decided, the state cites no statistical, experiential, or authoritative evidence to support this contention. It simply reargues the points considered in Neakok in hopes that the current court will disagree with the decision. This is precisely what the rule of stare decisis seeks to prevent: "The stare decisis doctrine rests on a solid bedrock of practicality: `no judicial system could do society's work if it eyed each issue afresh in every case that raised it.'"^[4]

Moreover, the point overturned by the court is not squarely raised here. The opinion overrules Neakok's statement that "[f]ormulation of [a] parole plan, and selection of special [parole] conditions"^[5] are non-immune operational functions.^[6] As set out in Neakok, this ruling addressed only a prison counselor's task of establishing a parole plan and a parole officer's authority to set special conditions of parole. Neither of these functions is at issue here.^[7]

The court nevertheless holds that Neakok's ruling on this point must be overturned to prevent it from applying to the parole board.^[8] Yet Neakok expressly declined to extend its immunity ruling to the parole board or its members, emphasizing that parole boards "have frequently been afforded quasi-judicial immunity from liability for their decisions."^[9] Since, by its own terms, Neakok did not apply this aspect of its ruling to the parole board's functions, there is no need to overturn the ruling in addressing the parole board's immunity here. Although the court may have well-founded concerns that Neakok's ruling might be extended to parole board functions in future cases, there is no need to overrule Neakok to resolve these concerns: the court can simply hold that Neakok's immunity ruling does not extend to the parole board.

My second reservation about today's opinion concerns its discussion of a parole officer's discretionary function immunity. In my view, the court's carefully targeted narrowing of Neakok's broad language discussing the scope of a parole officer's actionable duty leaves little need to further limit or refine our already well-defined law addressing discretionary function immunity. In this respect, while I certainly agree with the court's view that a mandatory DOC policy provides a sufficient basis for declaring a parole officer's function to be operational, I am concerned that the opinion might be read to say the converse, as well—that a non-mandatory policy necessarily suffices to establish a policy planning function.

I do not read the opinion as so holding. The court correctly recognizes that a DOC policy conferring discretion on parole officers may often prove determinative by signaling that the described function actually involves policy and planning. But a rule intractably declaring that all non-mandatory policies describe functions that are automatically shielded would run counter to our well-settled case law^[10] and could lead to anomalous results.^[11]

My last reservation about today's opinion concerns its observation addressing the companion case, C.J. v. State.^[12] Specifically, applying the narrowed duty analysis adopted here to C.J., the court states that in C.J. "it was foreseeable that if not properly supervised, the offender, a convicted rapist, might rape a woman in the community into which he was released."^[13] Although I have no doubt that this risk was foreseeable, I would hesitate to conclude that the entire population of a city like Anchorage constitutes a sufficiently identifiable "victim class" for purposes of establishing an actionable duty under Neakok. To the extent that the court's observation suggests otherwise, I would rely on the concurring view I express on this point in C.J.^[14]

MATTHEWS, Justice, dissenting in part.

I agree with the opinion authored by Justice Fabe insofar as it overrules Division of Corrections v. Neakok^[1] and holds that the discretionary function exception immunizes parole board release and revocation decisions. Similarly, I agree that discretionary immunity applies to parole officer decisions concerning conditions of parole and that Neakok must be overruled on this point as well. I also agree that a parole officer has no duty to seek out, without notice, possible parole violations, and I agree that parole officer decisions concerning parole revocation for minor violations of parole conditions are properly protected as discretionary. But I think the court should go further and consider whether Neakok correctly resolved the question of whether parole officers, and the State, generally should be subject to tort liability for negligent supervision of parolees.

In my view this question should be answered in the negative primarily because parole officers do not have sufficient control over parolees to be able to control their behavior. Further, imposing a general tort duty of supervision creates an incentive for parole officers to choose the most restrictive alternative available to them and may interfere with rehabilitative programs. I favor a rule that would subject parole officers to tort liability only when the officer has notice that his or her charge presents an imminent danger of harm to a particular person or class of persons. Such a rule would not be subject to the above deficiencies and would be consistent with our recent case law in analogous areas concerning torts involving the duty to control the actions of others. I discuss these points in the paragraphs that follow.

NEAKOK'S DUTY ANALYSIS SHOULD BE RE-EXAMINED.

The holding in Neakok rested principally on three of the factors we adopted in D.S.W. v. Fairbanks North Star Borough School District^[2] to aid in deciding whether a duty of care for tort law purposes should exist. According to the Neakok court, the foreseeability of the harm caused by the State's failure to use due care when supervising parolees, the closeness of the connection between that failure and the potential harm caused, and the insignificance of any burden imposed on the State and the community all weighed in favor of imposing a duty of due care on the State. I disagreed with the outcome then,^[3] and in my view new information shows that the Neakok court's conclusion was unsound.

Are Criminal Acts a Foreseeable Risk of a Failure to Use Due Care in Parole Supervision?

Foreseeability is said to be the most important D.S.W. factor. Neakok's conclusion that the State has a "substantial ability to control the parolee" is critical to its determination that the State can be held liable for the foreseeable harm caused by negligently supervised parolees.^[4] However, in practice, a parole officer has little or no meaningful control over the conduct of parolees. The facts here illustrate the point. Parole Officer Beckner is charged with incorrectly assigning McGrew to a "medium" instead of "maximum" supervision level. But even if McGrew had been subject to maximum supervision, she would have met with him in her office for a short time only twice monthly and would have three "field" visits per year.

Recidivism upon release from prison is, of course, foreseeable. According to a 2005 study published by the Urban Institute ("Study") "fewer than half of parolees successfully complete their period of parole supervision without violating a condition of release or committing a new offense, and . . . two-thirds of all prisoners are rearrested within three years of release."^[5] But the foreseeability of recidivism in the abstract is not the central concern of a duty analysis. The right question is whether parole supervision at today's levels reduces recidivism. The Study gives a generally negative answer to this question. It concludes that "the over-all effect of supervision appears to be minimal."^[6] With respect to violent offenders—the category with which tort law is most concerned—the Study's statistics show that the predicted probability of rearrest within two years after release is remarkably uniform regardless of supervision. The rate is fifty-five percent for those who are released unconditionally, fifty-six percent for mandatory parolees such as McGrew in this case, and fifty-five percent for discretionary parolees.^[7] Concerning offender types, the Study concludes that "only property offenders released to discretionary parole benefit from supervision. Violent offenders released to supervision are no less likely to be rearrested than their unsupervised counterparts."^[8]

The Study also underscores the point that parolee supervision as it is currently funded and conducted has little chance to control the conduct of parolees:

[x] Parole supervision is, in fact, quite minimal in most cases. Most parole officers manage large caseloads . . . and typically meet with individuals for about 15 minutes once or twice a month. Why would we expect such a small amount of contact to make a large amount of difference? Parolees don't: according to one study of parolees, most report that their parole officer did not have a major positive or negative impact on their postprison behavior. Clearly parole supervision must be more than occasional if it is to have an appreciable effect.^[9]

The Study notes that scholars have made a number of suggestions concerning changes in post-prison programs that might ultimately prove to be effective.^[10] But as to parole as it is currently conducted, the Study concludes that, "our reliance on parole serves little purpose apart from providing false comfort."^[11]

Reliance on the efficacy of parole is the central premise of Neakok's duty analysis. The court determined that the State stands in a "special relationship"^[12] to parolees "because of its substantial ability to control the parolee. Given this special relationship, it is not unreasonable to impose a duty of care on the state to protect the victims of parolees."^[13] Because the main premise of Neakok—the "substantial ability to control the parolee"—is invalid given parole supervision as currently designed and funded, Neakok's conclusion that this ability leads to a "special relationship" giving rise to a tort duty is also invalid.

Other states have recognized that parole officers do not have sufficient control of parolees to justify imposing a tort duty. The Supreme Court of Kansas in Schmidt v. HTG, Inc. held that parole officers lack sufficient control over parolees to create a special relationship between the officer and the parolee under which a duty to control the conduct of the parolee in tort might arise: "A parole officer does not take charge or exercise control over a parolee so as to . . . impos[e] a duty upon the State to control the conduct of the parolee to prevent harm to other persons or property."^[14] The high courts of Maryland, South Dakota, and Virginia have held likewise.^[15]

Closeness of the Connection

Another D.S.W. factor mentioned by Neakok is "the closeness of connection between the defendant's conduct and the plaintiff's injury."^[16]Neakok simply assumed that a parole officer's failure to supervise "can be viewed as closely connected to" an injury suffered at the hands of a parolee.^[17] Yet as the facts here illustrate, the relationship between a parole officer's allegedly negligent supervision and a parolee's criminal acts can be far from obvious. Today's opinion suggests that Beckner's acts of negligence in supervising McGrew included incorrectly calculating his risk score so that he was assigned a medium supervision level rather than a maximum supervision level.^[18] The more intensive level of supervision would have meant two monthly meetings rather than one and a field visit at least once every four months. These were errors that related to McGrew's parole between his release from prison in November 1996 and July 17, 1998, when Beckner filed a violation report that resulted in the issuance of an arrest warrant. How close was the connection between these alleged acts of negligence and the murder that McGrew committed on July 24, 1999? Given that at the time of the murder McGrew had been arrested and released by a discretionary act of a parole board member, and his revocation hearing was pending, one can characterize the closeness of the connection between these acts and the murder as either extremely remote or non-existent.^[19]

I also observe, albeit repetitively because the D.S.W. factors overlap, that the idea that there is a close connection between supervision of a parolee and preventing a parolee's criminal activities is also belied by the statistics contained in the Urban Institute Study referenced above. These statistics indicate that recidivism is not reduced by parole supervision for parolees convicted of violent crimes. The Study tells us that violent offenders who are released and not supervised do not re-offend at a higher rate than violent offenders who are released and supervised. In other words, parole as it is currently practiced and funded has a minimal effect on public safety. Supervision, whether negligent or not, therefore has little bearing on whether an offender will commit new crimes. Thus, again, the idea that the relationship between the supervision of a parolee and a parolee's crimes is close enough to justify tort liability is, in my opinion, untenable.

Burdens and Consequences

Neakok dismissed the notion that its holding imposed significant burdens on the State and community by focusing primarily on the financial impact of tort liability. Potential financial impact is an important factor given the recidivism rate of parolees and the frequency of parole violations. But a potentially more important burden is the effect on the parole system.

In State v. Sandsness we adopted the Vermont Supreme Court's reasoning in Sorge v. State that the State owes no duty to control juvenile offenders that it releases from custody.^[20] We accepted the Sorge court's reasoning that because the focus of juvenile jurisdiction is rehabilitation, the State could not be liable to third persons harmed by juveniles negligently released from State rehabilitative programs. The imposition of a duty to control under tort law in those circumstances would cause the State to err on the side of detention and thus interfere with the rehabilitative process.^[21] Rehabilitation is also an important function of parole. Therefore, while the facts here are distinguishable from those in Sandsness and Sorge, the effect on parole-related programs is similar.

Today's opinion suggests that the rehabilitative function of parole is preserved because discretionary immunity shields the State from liability for negligently releasing parolees from detention.^[22] But the prospect of being sued for negligent supervision once a parolee is released may cause the State to err on the side of continued detention. In addition, once a parolee is released, either as a matter of discretion, or as a matter of right, as in this case, the prospect of negligent supervision liability will have a tendency to cause the State to err on the side of renewed confinement when violations are suspected or discovered.

The Sorge court reviewed the authorities that had rejected liability because of the potential interference that it could have on post-prison release programs at some length:

Furthermore, this application of the § 319 exception to the rule that there is no common-law duty to control the conduct of another to protect a third person has been rejected by jurisdictions that have recognized that most juvenile and adult programs dealing with persons committed to the custody of the State are intended to rehabilitate conduct rather than control it. See Davenport v. Community Corrections of Pikes Peak Region, Inc., 962 P.2d 963, 968 (Colo.1998) ("Community corrections . . . programs designed to . . . reintegrate incarcerated offenders into society."); Finnegan v. State, 138 Vt. 603, 420 A.2d 104 (1980); see also Ruf v. Honolulu Police Dep't, 89 Hawai`i 315, 972 P.2d 1081, 1093 (1999) (in declining to apply § 319 to alleged negligent release of prisoner by police, court notes that risk of liability could pressure police to "err more often on the side of excessive detention"). Thus, for example, parents of a restaurant employee who was raped and killed by a coworker, who had been conditionally released from prison, asserted that § 319 of the Restatement established a duty of care on the part of the Kansas Department of Corrections. The Supreme Court of Kansas found that neither the parole officer nor the department of corrections had charge of the individual who committed the assault to the extent necessary to fall within § 319. See Schmidt v. HTG, Inc., 265 Kan. 372, 961 P.2d 677, 687 (1998). In rejecting the imposition of liability upon the State under the rationale of § 319, the court observed:

[An] overbroad construction . . . escalates the State's responsibility to that of the virtual guarantor of the safety of each and every one of its citizens from illegal and unlawful actions of every parolee or person released from custody under any type or kind of supervision.

Id.; see also Thompson v. County of Alameda, 27 Cal.3d 741, 167 Cal.Rptr. 70, 614 P.2d 728, 735 (1980) ("parole and probation release . . . comprise an integral and continuing part in our correctional system authorized by the Legislature, serving the public by rehabilitating substantial numbers of offenders and returning them to a productive position in society."); Whitcombe v. County of Yolo, 73 Cal. App.3d 698, 141 Cal.Rptr. 189, 199 (1977) ("Were we to find a cause of action [here] we would in effect be encouraging the detention of prisoners in disregard of their rights and society's needs."); Rivers, 133 Vt. at 14, 328 A.2d at 400 (argument that State should be liable for harm caused by person on rehabilitative release for failing to control such persons "runs dangerously parallel to the arguments for preventative detention").^[23]

Based on this review, I do not think that there is a case to be made that tort liability should be imposed for negligent supervision of parolees. To summarize, parole supervision of violent offenders has been shown to be ineffective in preventing them from committing new crimes. Since parole as presently conducted is ineffective at controlling the conduct of parolees, no special relationship giving rise to a tort duty to control them should arise. Further, it cannot be said that there is typically a close connection between acts of negligent supervision and the criminal conduct of parolees.

But two things are clear. First, acts of recidivism are frequent, and the imposition of tort liability for parole supervision shortcomings will have a tendency to skew decision-making concerning parolees in favor of the most restrictive choice available. Second, the premise on which Neakok based its conclusion that parole officers have a tort duty to control parolees—that parole supervision is effective in reducing recidivism in violent offenders—was based on an unexamined assumption that the best information available today shows to be incorrect. Thus the conditions for overruling a prior decision of this court are satisfied: "A prior decision should be overruled only if the court is clearly convinced that the precedent is erroneous or no longer sound because of changed conditions, and that more good than harm would result from overturning the case."^[24]

LIABILITY WHERE THERE IS NOTICE OF IMMINENT PERIL.

I do not suggest that parole officers should, in all circumstances, be immune from tort liability. If a parole officer has knowledge of a specific threat of imminent harm to a person or class of persons, the parole officer should have a tort duty to take appropriate action. Such a rule would bring the tort duties of parole officers in line with parents and police officers. A parent, for example, owes a duty to restrain a child only when there is reason for the parent to know that the child poses an imminent and foreseeable risk of harm.^[25] Similarly, police officers owe "a duty of reasonable care . . . to respond to threats of imminent, life-threatening, assaultive conduct when given sufficient specific information to respond."^[26]

For these reasons, I respectfully dissent in part from Justice Fabe's opinion.

NOTES

[1] 721 P.2d 1121, 1125 (Alaska 1986).

[2] See id. at 1134. Our decision in Neakok addressed this question only in the context of whether a parole officer could be held liable for decisions regarding selection of parole conditions because we concluded that no reasonable jury could find the board at fault in that case. Id. at 1125 n. 4. Whether discretionary function immunity applies to decisions made by the parole board, such as the selection of parole conditions and failure to revoke parole, as well as decisions made by the parole officer, is at issue in this case.

[3] Because this case comes to us from a denial of summary judgment, the statement of facts reflects the respondent's version of the disputed facts. See Gordon v. Alaska Pacific Bancorporation, 753 P.2d 721, 722 n. 1 (Alaska 1988).

[4] McGrew also received a consecutive sentence of three years, with two and one-half years suspended, for escape in the second degree, as well as suspended sentences for theft and criminal mischief. He was also ordered to serve three years of probation following his incarceration.

[5] DOC Policy 902.03 (1988). Because both parties cite the DOC Policy dated July 14, 1988, we assume that this policy reflects the provisions in force at all relevant times for this case.

[6] Id.

[7] Id.

[8] According to Beckner, Darryl Poindexter, who had recently been arrested, had at one time used McGrew's identification card, causing the ongoing confusion about identity.

[9] State, Dep't of Health & Soc. Servs. v. Sandsness, 72 P.3d 299, 301 (Alaska 2003).

[10] Id.

[11] Kiokun v. State, Dep't of Pub. Safety, 74 P.3d 209, 212 (Alaska 2003).

[12] State v. Coon, 974 P.2d 386, 394 (Alaska 1999) (quoting State v. Fremgen, 914 P.2d 1244, 1245 (Alaska 1996)).

[13] 721 P.2d at 1126.

[14] Id. at 1124.

[15] Id. at 1125, 1134.

[16] See Guerrero v. Alaska Hous. Fin. Corp., 6 P.3d 250, 254 (Alaska 2000).

[17] Kiokun, 74 P.3d at 213 (proceeding directly to immunity because that analysis "illustrates the public policy issues that would also bear on a duty analysis").

[18] AS 09.50.250(1).

[19] Angnabooguk v. State, Dep't of Natural Res., 26 P.3d 447, 453 (Alaska 2001) (quoting State, Dep't of Transp. & Pub. Facilities v. Sanders, 944 P.2d 453, 456 (Alaska 1997)).

[20] Id.

[21] Id.

[22] Cf. Estate of Arrowwood v. State, 894 P.2d 642, 645-46 (Alaska 1995) (holding that officials' decision to keep road open could not be described as merely operational where "relevant statutory and administrative code provisions do not require officials to act to carry out a predetermined policy" and "delegate[ ] to officials on the scene the authority to act" based on their evaluation of conditions).

[23] Id.

[24] Kiokun, 74 P.3d at 215.

[25] See Angnabooguk, 26 P.3d at 455 (stating that "we have never held that an entire class of decisions . . . are necessarily bound up with policy considerations").

[26] See id. at 458.

[27] Cf. id. (declining to decide whether each allegation in the complaint concerns planning or operational decisions and instructing superior court on remand to make a separate determination for each allegation following the principles announced by the court).

[28] The parole board generally imposes the conditions of parole, including special conditions such as therapy or treatment. AS 33.16.150. Parole officers may also impose special conditions of parole. It is unclear whether the complaint alleges that the parole officer as well as the parole board failed to impose appropriate parole conditions on McGrew. Our conclusion that decisions regarding parole conditions are discretionary applies to decisions made by the parole officer as well as the parole board.

[29] The parole board determines whether to revoke an offender's parole. AS 33.16.220.

[30] 721 P.2d at 1137 (Matthews, J., dissenting).

[31] Cf. Kiokun, 74 P.3d at 216 (finding that state troopers' decision whether to launch a search and rescue effort was immune because it was based on resource allocation and public policy considerations).

[32] Estate of Arrowwood v. State, 894 P.2d 642, 645 (Alaska 1995). Discretionary function immunity also ensures that courts "avoid the re-examination of decisions which lie outside the realm of their institutional competence." Id.

[33] 721 P.2d at 1134.

[34] See id. at 1125 n. 4.

[35] Sandsness, 72 P.3d at 308 (Matthews, J., dissenting).

[36] Cf. Coon, 974 P.2d at 394 (discussing the standard for overruling a prior decision of this court).

[37] DOC Policy 902.14 (1988).

[38] Id.

[39] Id. The administrative regulation regarding reporting of parole violations similarly provides that all felony behavior and any "serious misdemeanor behavior" indicating that the parolee is a danger to the public "must be reported to the board." 22 AAC 20.350(a)-(b). The regulation further provides that if a parolee fails to report once, the parole officer "may notify the board," but the parole officer "shall" notify the board if the parolee misses two consecutive reporting periods. 22 AAC 20.350(c). In the case of any other type of parole violation, "[r]eporting of the violation may be held in abeyance in the discretion of the parole officer." 22 AAC 20.350(d).

[40] See Angnabooguk, 26 P.3d at 453.

[41] She did, however, have notice that the police had come to McGrew's house to arrest him. We address below whether she had a duty to investigate based on this information. Cowles argues that DOC policy required Beckner to file a violation report and recommend full or partial revocation of parole because some of McGrew's violations involved positive tests for marijuana. As the State points out, DOC Policy 902.25 only requires a parole officer to recommend partial or full revocation in response to a positive test if the parolee also committed one or more serious violations or is deemed to be a high risk to the community. Use of marijuana is a class B misdemeanor and therefore does not qualify as a serious violation. AS 11.71.060; AS 11.71.190.

[42] DOC Policy 902.03.

[43] DOC Policy 902.14.

[44] See Neakok, 721 P.2d at 1134 (noting that parole officer's actions in supervising offenders are ministerial in nature).

[45] See, e.g., Acevedo by Acevedo v. Pima County Adult Probation Dep't, 142 Ariz. 319, 690 P.2d 38, 41 (1984) (holding that probation officers who allowed defendant to have contact with minors in violation of court order could not assert sovereign immunity); A.L. v. Commonwealth, 402 Mass. 234, 521 N.E.2d 1017, 1024 (1988) ("A probation officer's duty to make reasonable efforts to ascertain whether a probationer has complied with the terms of his or her probation is not a discretionary function."); Taggart v. State, 118 Wash.2d 195, 822 P.2d 243, 252 (1992) (holding that discretionary function immunity does not shield parole supervision decisions).

[46] Cf. Adams v. City of Tenakee Springs, 963 P.2d 1047, 1051 (Alaska 1998) ("Decisions about how to allocate scarce resources are matters of policy immune from judicial review.").

[47] Cf. Kiokun, 74 P.3d at 217 (holding that state troopers' decision whether to undertake a search and rescue operation was immune because it involved balancing of "public safety objectives and the allocation of resources").

[48] 721 P.2d at 1137 (Matthews, J., dissenting).

[49] 72 P.3d at 304.

[50] Id. at 300.

[51] Id.

[52] Id.

[53] Id. at 304 (quoting Sorge v. State, 171 Vt. 171, 762 A.2d 816, 821 (2000)).

[54] Id. at 301.

[55] 171 Vt. 171, 762 A.2d 816 (2000).

[56] Id. at 821.

[57] Coon, 974 P.2d at 394.

[58] RESTATEMENT (THIRD) OF TORTS § 41 cmt. f (Proposed Final Draft No. 1, 2005).

[59] See, e.g., Starkenburg v. State, 282 Mont. 1, 934 P.2d 1018, 1028 (1997); Faile v. South Carolina Dep't of Juvenile Justice, 350 S.C. 315, 566 S.E.2d 536 (2002); Hertog ex rel. S.A.H. v. City of Seattle, 138 Wash.2d 265, 979 P.2d 400, 409 (1999).

[60] A.L. v. Commonwealth, 402 Mass. 234, 521 N.E.2d 1017, 1021 (1998).

[61] Id.

[62] Id.; see also Sandsness, 72 P.3d at 305 (stating that foreseeability is "the most important single D.S.W. factor"); D.S.W. v. Fairbanks North Star Borough Sch. Dist., 628 P.2d 554, 555 (Alaska 1981) (noting that foreseeability is a factor in determining whether to impose a duty of care).

[63] 721 P.2d at 1129.

[64] Id.

[65] 151 P.3d 373, 2006 WL 3692501, Op. No. 6081 (Alaska, December 15, 2006).

[66] 721 P.2d at 1138 (Matthews, J., dissenting) (citing City of Kotzebue v. McLean, 702 P.2d 1309 (Alaska 1985)).

[67] RESTATEMENT (THIRD) OF TORTS § 41 cmt. f (Proposed Final Draft No. 1, 2005).

[68] 972 P.2d 978, 986 (Alaska 1999).

[69] Dissent at 371.

[70] Cf. City of Kotzebue v. McLean, 702 P.2d 1309, 1315 (Alaska 1985) (reasoning that imposing duty on police officers to respond to imminent threats was consistent with officers' training and created an incentive for officers to follow their own policies and procedures).

[71] See D.S.W., 628 P.2d 554.

[72] Waskey v. Municipality of Anchorage, 909 P.2d 342, 343-44 (Alaska 1996) (noting that it was "unnecessary to resort to the D.S.W. approach" where the court had previously "decided two cases more closely related to this case"). We did apply the D.S.W. factors in Sandsness. 72 P.3d at 305. But there we explicitly held that Neakok did not control, making applicable the D.S.W. framework. Id.

[73] Sharp v. Fairbanks N. Star Borough, 569 P.2d 178, 181 (Alaska 1977) (internal quotations omitted).

[74] Id. (quoting State v. Abbott, 498 P.2d 712, 727 (Alaska 1972)).

[1] State, Div. of Corr. v. Neakok, 721 P.2d 1121 (Alaska 1986).

[2] Thomas v. Anchorage Equal Rights Comm'n, 102 P.3d 937, 943 (Alaska 2004) (quoting State, Commercial Fisheries Entry Comm'n v. Carlson, 65 P.3d 851, 859 (Alaska 2003) (internal quotations omitted)).

[3] In his dissenting opinion, Justice Matthews cites parole statistics from a recent multi-state study by the Urban Institute—Amy Solomon, Vera Kachnowski & Avinash Bhati, Does Parole Work? (Urban Institute, March 2005) (hereinafter "study")—as a basis to argue that Neakok should be overruled because statistical evidence now disproves Neakok's main premise—that parole officers have a "substantial ability to control the parolee." Dissent at 370. But the study's statistics do not support the dissent's theory. The study—which, incidentally, excluded Alaska— compiled data from fifteen different systems without taking into account variations potentially attributable to funding, staffing, training, or types of supervision within or among the systems studied. Based on these statistics, the study merely compared rearrest rates for prisoners released with and without supervision in the selected locations. As its title suggests, the study was designed as a descriptive measure, not a predictive one: it merely asked "Does Parole Work" as it currently exists—not "Could Parole Work" in a better-run system. Moreover, because the study did not consider whether the states it studied imposed civil liability for negligent parole supervision, its findings shed no light on whether or how civil liability might affect parole officers' supervision of their parolees. Last, and most notably, the authors of the study expressly acknowledged these limitations, warning the study's readers that its statistical findings could not support sweeping generalizations like the dissent makes here:

It bears repeating that the nature of our analysis does not allow for insights into whether certain types of supervision, such as neighborhood-based or case management models, are more effective than others or whether there are differences in outcomes across states. It is also unclear how much rearrest outcomes are the result of policy directives (e.g., a decision to watch more closely and arrest more quickly) and not criminal activity alone.

Study at 15.

[4] Thomas, 102 P.3d at 943 (quoting Planned Parenthood of Southeastern Pennsylvania v. Casey, 505 U.S. 833, 854, 112 S.Ct. 2791, 120 L.Ed.2d 674 (1992)).

[5] Neakok, 721 P.2d at 1134.

[6] Maj. Op. at 360.

[7] There is no hint of any claim against prison counselors here, nor is there any claim that McGrew's parole officer, Beckner, had any authority to set special conditions of parole. Notably, under current law, parole officers have no authority to set special conditions of parole unless the authority has been actually delegated by the parole board or a board member. AS 33.16.150(c).

[8] Maj. Op. at 360.

[9] Neakok, 721 P.2d at 1133 n. 19 (citations omitted).

[10] See, e.g., State v. Abbott, 498 P.2d 712, 720-22 (Alaska 1972).

[11] Specifically, a rule that inflexibly viewed all non-mandatory policies as describing immunized functions would leave the department in charge of its own immunity by allowing it to make all policies non-mandatory, thereby converting all functions performed by parole officers, "even driving a nail," id., into planning and policy decisions.

[12] C.J. v. State, Dep't of Corr., 151 P.3d 373, 2006 WL 3692501, Op. No. 6081 (Alaska, December 15, 2006).

[13] Maj. Op. at 363-64.

[14] C.J., 151 P.3d at 384-85.

[1] 721 P.2d 1121 (Alaska 1986).

[2] 628 P.2d 554 (Alaska 1981).

[3] Neakok, 721 P.2d at 1137 (Matthews, J., dissenting).

[4] Id. at 1126.

[5] Amy L. Solomon et al., Does Parole Work? URBAN INSTITUTE, 1 (2005), available at http://www. urban.org/UploadedPDF/311156—Does—Parole—Work.pdf. The purpose of this study was "to assess, at an aggregate level, whether parole `works' at reducing recidivism among those who are supervised after release from state prison." Id. at 1.

[6] Id. at 15.

[7] Id. at 10, table 3.

[8] Id. at 2. I set out here in full the Research Highlights of the Study:

[x] Overall, parole supervision has little effect on rearrest rates of released prisoners. Mandatory parolees, who account for the largest share of released prisoners, fare no better on supervision than similar prisoners released without supervision. In fact, in some cases they fare worse. While discretionary parolees are less likely to be rearrested, this difference narrows (to 4 percentage points) after taking into account personal characteristics and criminal histories.

[x] Certain prisoners benefit more from supervision—especially discretionary release to supervision—than others. For example, females, individuals with few prior arrests, public order offenders, and technical violators are less likely to be rearrested if supervised after prison. Persons with a combination of these characteristics, representing relatively low-level offenders, exhibit even lower rearrest rates if supervised. Conversely, supervision does not improve rearrest outcomes for some of the higher rate, more serious offenders.

[x] Of the largest groups of released prisoners—male drug, property, and violent offenders—only property offenders released to discretionary parole benefit from supervision. Violent offenders released to supervision are no less likely to be rearrested than their unsupervised counterparts. For male drug offenders, mandatory release to supervision predicts higher rearrest rates than for unconditional releasees or discretionary parolees.

Id. at 1-2 (emphasis added).

[9] Id. at 16 (footnotes omitted). The description in this paragraph of the typical level of parole supervision squares with Alaska parole supervision as currently established. See infra page 370-71 (Maximum supervision entails two monthly meetings and a field visit once every four months.).

[10] Id. at 16-17

[11] Id. at 17. Chief Justice Bryner criticizes my reliance on the Study. He suggests that the Study is merely descriptive of the effectiveness of parole in the locations studied and does not show that parole could not be effective in better run systems. Further, he claims that the data relied upon by the Study is too limited to be relied on in Alaska since it is based on statistics from only fifteen states, not including Alaska. He also argues that the limitations inherent in the Study mean that generalizations about the ineffectiveness of parole supervision cannot be drawn. With respect to this point he quotes a caution set out in the Study relating to certain types of supervision. Maj. Op. at 366 note 3 (Concurring Opinion of Chief Justice Bryner).

My response to these points follows. The Study clearly concludes that supervised parole as it is traditionally practiced in the United States with respect to violent offenders is ineffective at reducing recidivism rates. The recidivism rates for violent offenders who are subject to supervision are, at best, only equal to the rates for violent offenders who are not subject to supervised parole. They are not lower as might be expected. The Study, of course, does not attempt to isolate any role that negligent supervision might have on outcomes. But unless we are to assume that all, or nearly all, of the supervision reflected in the Study was negligent, the Study does indicate that non-negligent supervision does not produce better results than no supervision at all. Otherwise, a reduction in recidivism rates would be reflected.

It is of course possible that if parole supervision programs were designed differently, or if substantially more money were committed to parole supervision, outcomes might improve. But in deciding duty questions for tort law purposes we must look at programs as they exist and ask whether it makes sense to impose tort law liability for program failures. For example, in D.S.W. the question was whether there should be a tort duty imposed on public schools to recognize and treat learning disabilities. We held that there should not be such a duty, citing, among other reasons, the numerous "social and financial problems" with which the public schools are beset. 628 P.2d at 556. If there is one thing clear in this case, it is that decisions as to the design and funding level of parole supervision are not reviewable in tort. Thus, the fact that different programs might be effective cannot be the basis for imposing liability for the shortcomings of the current program.

Chief Justice Bryner's criticism that the database used by the Study—results from fifteen states—is too limited, is also unwarranted. Maj. Op. at 366 note 3. The scholars who conducted the Study thought the data was sufficient to draw the conclusions that they reached, and the focus of the Study was on "national-level trends." Study at 14.

Chief Justice Bryner's point that the Study does not support the generalizations that I have relied on in this dissent also lacks validity. The generalizations that I rely on are those that the authors of the Study have made. Specifically, the Study states: "Mandatory parolees, who account for the largest share of released prisoners, fare no better on supervision than similar prisoners released without supervision." Study at 1. "Violent offenders released to supervision are no less likely to be rearrested than their unsupervised counterparts." Id. at 2. With respect to the language that Chief Justice Bryner quotes, the scholarly caution reflected there does not, in the view of the authors, undermine their overall conclusion. Indeed, the sentence immediately following the language quoted by the Chief Justice states: "At the same time, given our country's heavy reliance on parole to manage those released from prison, it is discouraging—although not wholly unexpected—to find that the overall effect of supervision appears to be minimal." Id. at 15.

[12] That is, a relationship giving rise to a duty to control the conduct of a third person. See RESTATEMENT (SECOND) OF TORTS § 315(a) (1965).

[13] Neakok, 721 P.2d at 1126-27 (emphasis added).

[14] 265 Kan. 372, 961 P.2d 677, 679 (1998).

[15] Lamb v. Hopkins, 303 Md. 236, 492 A.2d 1297, 1304 (1985); Small v. McKennan Hosp., 403 N.W.2d 410, 414 (S.D.1987); Fox v. Custis, 236 Va. 69, 372 S.E.2d 373, 376 (1988).

[16] 721 P.2d at 1125.

[17] Id. at 1129.

[18] Maj. Op. at 360, 361-62 (Opinion of Justice Fabe).

[19] Further, as the facts of the companion to this case, C.J. v. State, 151 P.3d 373, 2006 WL 3692501, Op. No. 6081 (Alaska, December 15, 2006), indicate, issuing a parole violation report and an arrest warrant may not suffice to prevent a parolee from committing new crimes.

[20] State v. Sandsness, 72 P.3d 299, 302-03 (Alaska 2003) (citing Sorge v. State, 171 Vt. 171, 762 A.2d 816 (2000)).

[21] Id. at 302.

[22] Maj. Op. at 364 (Opinion of Justice Fabe).

[23] 762 A.2d at 820-21.

[24] Kinegak v. State, Dep't of Corrections, 129 P.3d 887, 889-90 (Alaska 2006).

[25] Dinsmore-Poff v. Alvord, 972 P.2d 978, 986 (Alaska 1999) (holding that a parent's tort duty to restrain child exists only where "the parent ha[s] reason to know with some specificity of a present opportunity and need to restrain the child to prevent some imminently foreseeable harm").

[26] Dore v. City of Fairbanks, 31 P.3d 788, 795 (Alaska 2001) (discussing the meaning of City of Kotzebue v. McLean, 702 P.2d 1309 (Alaska 1985)).