Opinion by Judge FISHER; Dissent by Judge O’SCANNLAIN.
OPINION
FISHER, Circuit Judge:We consider whether Douglas R. Stankewitz received ineffective assistance of counsel at the penalty phase of his capital murder trial. In a prior appeal in this matter, we held that Stankewitz’s allega*1165tions that his counsel failed to investigate and present readily available mitigation evidence — including evidence of his deprived and abusive upbringing, potential mental illness, long history of substance abuse and use of substantial quantities of drugs leading up to the murder — if true, would establish that he received ineffective assistance. We remanded for an evidentiary hearing so that the state would have an opportunity to challenge Stankewitz’s allegations. On remand, however, the state agreed to proceed without an evidentiary hearing. The district court, after considering several thousands of pages of documents describing Stankewitz’s troubled background, found that the state had failed to rebut most of Stankewitz’s allegations. The court therefore held that Stankewitz’s counsel’s performance fell below the constitutional standards articulated in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), and granted his petition for a writ of habeas corpus. We affirm.
I.
We recounted the factual and procedural history in our prior opinion, see Stankewitz v. Woodford, 365 F.3d 706, 708-12 (9th Cir.2004), and do not repeat it in detail here. In brief, Stankewitz was convicted and sentenced to death in 1978 for the murder of Theresa Greybeal. The California Supreme Court reversed that conviction upon automatic appeal because the trial court failed to address a conflict between Stankewitz and his public defender, Salvatore Sciandra. Before Stankewitz’s second trial, the trial court found that a conflict indeed existed between Sciandra and Stankewitz and appointed private counsel, Hugh Goodwin. The jury ultimately convicted Stankewitz and again sentenced him to death. After the California Supreme Court rejected Stankewitz’s state postconviction challenges, he filed the present habeas petition in federal court, raising several challenges to the guilt and penalty phases of his trial. The district court denied the petition in its entirety without holding an evidentiary hearing as to any of Stankewitz’s claims.
We affirmed the district court with respect to Stankewitz’s guilt-phase challenges. See Stankewitz v. Woodford, 94 FedAppx. 600 (9th Cir.2004) (unpublished). We reversed, however, as to Stankewitz’s claim that Goodwin rendered ineffective assistance of counsel during the penalty phase of Stankewitz’s trial by failing to investigate and present available mitigating evidence. See Stankewitz, 365 F.3d 706.
With respect to that claim, we undertook a detailed review of the mitigating circumstances Stankewitz alleged, Goodwin’s investigation and penalty phase performance and the totality of the evidence introduced at trial, and held that Stankewitz’s allegations, if true, established that Goodwin’s performance was both deficient and prejudicial under Strickland. We observed that Goodwin’s penalty phase presentation was “minimal, consisting of testimony from six witnesses (only four of whom were actually in court) and covering only approximately 50 pages in the transcript.” Id. at 716. Two witnesses focused only on the “power of God” to help persons change their lives and provided no specific information about Stankewitz (aside from one witness’ admission that he had no reason to believe that Stankewitz was religious). See id. A third witness, by stipulation, testified only that he admired the work of prison chaplains. See id. at 716 & n. 4. We described this strategy as one that “had little hope of succeeding, and indeed seemed predestined to fail.” Id. at 716. From the remaining three witnesses, Goodwin elicited only vague references to Stankewitz’s history: the observation of sores and needle marks on his arm the day after the shoot*1166ing, one beating he received as a child, his placement in foster homes and the difficulties encountered on Indian reservations. See id. at 716-17. We noted that Goodwin elicited this testimony “in a cursory manner that was not particularly useful or compelling.” Id. at 724 (quoting Douglas v. Woodford, 316 F.3d 1079, 1090 (9th Cir.2003)). Goodwin also focused little on the actual details of Stankewitz’s life during his closing argument. See id. at 717.
We also observed that Goodwin failed to conduct even the most basic investigation of Stankewitz’s background. Goodwin never hired an investigator or interviewed Stankewitz’s teachers, foster parents, psychiatrists, psychologists or anyone else who may have examined or spent time with Stankewitz during his upbringing. See id. at 719. He did not interview anyone involved in Stankewitz’s first trial and thus did not know about the existence of any diagnoses relating to Stankewitz’s mental capabilities. See id. He did not procure a psychological examination of Stankewitz, even though he believed that Stankewitz was not mentally competent. See id. Furthermore, the six witnesses who did testify at the penalty phase were obtained with little effort. Stankewitz’s sister-in-law, for instance, became a witness because of a chance meeting with Goodwin in the courthouse. See id. at 720-21. Another witness merely had her testimony from Stankewitz’s first trial read into evidence. See id. at 720. Goodwin’s key witness, Probation Officer Joe Walden, met Stankewitz only once when Stankewitz was six years old and affirmed that Goodwin did nothing to help him prepare to testify. See id. at 724. The remaining three witnesses, two of whom provided no testimony specific to Stankewitz and one of whom provided only a two-sentence stipulation regarding the work of jail chaplains, advanced Goodwin’s apparent interest in the power of religion, but provided no mitigating information about Stankewitz. See id. at 711-12, 716 n. 4, 721. Goodwin also failed to investigate and rebut the prosecution’s aggravating evidence. See id. at 720 (describing Goodwin’s failure to investigate or rebut the prosecution’s testimony indicating that Stankewitz shot a police officer, despite readily available evidence that would have undermined the prosecution’s argument).
Finally, we observed that, in comparison to the meager mitigation evidence that Goodwin presented to the jury, Stankewitz made compelling allegations in his habeas petition regarding his deprived and abusive upbringing, potential mental illness, long history of drug use and consumption of substantial quantities of drugs in the days leading up to Greybeal’s murder. See id. at 717-19.
Based on those circumstances, we held that Stankewitz’s allegations, if true, would establish that he received ineffective assistance at his penalty phase proceeding. See id. at 722. We remanded for an evidentiary hearing so that “the state [would] have the opportunity to challenge Stankewitz’s allegations.” Id. at 725.
On remand, the district court expanded the record to include the files of the public defender in Stankewitz’s first trial, Seiandra, and many other documents proffered by Stankewitz. See Rules Governing Section 2254 Cases 7 (permitting the district court to expand the record). In total, several thousands of pages of documents were added to the record, including many reports by probation officers and other employees at juvenile institutions, psychological evaluations and declarations by family members and others close to Stankewitz. The parties then agreed to brief the merits based on the evidence in the record. Stankewitz argued that he was entitled to relief based on the documentary evidence, but, in the alternative, requested an evidentiary hearing to resolve any con*1167tested facts that precluded relief. The state took the position that no evidentiary hearing was necessary and that Stankewitz’s petition should be denied.
In September 2009, the district court issued an order granting Stankewitz a writ of habeas corpus. The court credited most of Stankewitz’s allegations, noting that many were proved by official documents in the record. The court found that “[e]ven accepting the Warden’s objections to some of Stankewitz’s allegations, the evidence shows Stankewitz was already severely emotionally damaged by the time he was removed from his home at age six.” Furthermore, Stankewitz’s evidence reflected “a deprived background, being institutionalized early in his life and essentially raised in institutions” and that Stankewitz “was hardened by the years of criminal associations and surroundings.” Relying on a social evaluation conducted when Stankewitz was nineteen, the court found that
[f]rom an early life developmental standpoint, Stankewitz has suffered from early childhood losses, prolonged separation from parents, poor institutional surrogate care. This has resulted in poor social adjustment as manifested by frequent runaways, behavior problems, scholastic under-achievement and finally culminating in anti-social behavior which has occurred both in and out of institutional placements.
Furthermore, although it found that some of Stankewitz’s allegations relating to childhood abuse had limited support, the court found that “the record as a whole shows Stankewitz was psychologically and emotionally damaged by his upbringing.” The court also observed that Stankewitz had a very severe substance abuse problem that began at age 10, and that he had binged on substantial quantities of alcohol, heroin and methamphetamine leading up to the murder.
The court denied the state’s motion for reconsideration. The state now appeals. It challenges both the district court’s findings of fact and the legal conclusion that Goodwin rendered ineffective assistance, and urges us to deny Stankewitz’s habeas petition or, in the alternative, remand again for an evidentiary hearing.
II.
We recognized in our prior decision, and the parties agree, that the Anti-terrorism and Effective Death Penalty Act of 1996 (AEDPA) does not apply to Stankewitz’s ineffective assistance claim. See Stankewitz, 365 F.3d at 713. We review Stankewitz’s claim under the standard articulated in Strickland, 466 U.S. 668, 104 S.Ct. 2052, which requires that Stankewitz show Goodwin’s representation was both deficient and prejudicial. See id. at 687, 104 S.Ct. 2052. We review de novo the legal question of whether Stankewitz received ineffective assistance of counsel and review the district court’s factual findings for clear error. See Stankewitz, 365 F.3d at 714.
III.
As an initial matter, the state challenges some of the district court’s findings regarding Stankewitz’s background. We begin by addressing these challenges, but hold that each of the court’s findings was adequately supported by the record.
First, the state challenges the district court’s conclusion that Stankewitz was severely emotionally damaged by his upbringing. The state acknowledges that Stankewitz was born into a dysfunctional family but argues that Stankewitz could not have been damaged because he spent very little time with his family after he was removed from his home at the age of six. *1168We have no trouble concluding that the district court did not clearly err by concluding that Stankewitz was severely damaged by his upbringing. We described many of Stankewitz’s allegations pertaining to “his difficult and traumatic youth” in our prior opinion. Stankewitz, 365 F.3d at 717-18. As the district court found, most of these allegations have been substantiated by documentary evidence added to the record. The documents illustrate that Stankewitz was born into a poverty-stricken home described by police and probation reports as dirty, covered in cockroaches and fleas, and without electricity or running water. There was often not enough food for Stankewitz and his nine siblings, who were “highly neglected.” A psychiatric evaluation of Stankewitz’s mother, Marian, confirms that she had been an alcoholic since she was a child and that she was severely intellectually impaired. Marian was arrested several times for crimes that include assault, grand theft auto and drunk driving, and she was ultimately convicted of voluntary manslaughter for shooting and killing a man while she was drunk at a party. According to Marian, after she got married, she would regularly drink three to four six packs of beer or two fifths of a gallon of whiskey in a night, including while she was pregnant with Stankewitz. A probation report described Stankewitz’s mother as “incapable of caring for herself and all of her children and certainly incapable of caring for Doug.” Stankewitz’s father, Robert, was an alcoholic truck driver and leader of a motorcycle gang. According to his rap sheet, he was arrested several times between 1951 and 1968 for crimes that include wife beating, robbery, non-support, public drunkenness, forgery, disturbing the peace and contributing to the delinquency of a minor. A judge described Robert as an “outlaw” and “a definite menace to society” who had “low intelligence,” was “without education,” had “no respect for the rights or feelings of other [sic]” and “like[d] violence.” According to Marian, Robert severely beat her while she was pregnant with Stankewitz, knocking her to the ground, kicking her stomach several times and breaking her nose. After Stankewitz was born, he and his siblings witnessed Robert beat and threaten to kill Marian and attempt to run her over with a car. On another occasion, Robert pulled a gun on Marian and fired several shots between her legs. Robert and Marian separated in 1966, when Stankewitz was eight, because of Robert’s brutal attacks.
According to Stankewitz’s sister and aunt, both of Stankewitz’s parents regularly beat all of their children. Robert often whipped them with a belt. On one occasion he came into the house in the middle of the night with a gun and threatened to shoot one of Stankewitz’s brothers. Marian often used electric cords or belts, and once even pulled a gun on Stankewitz’s sister. They beat the children more if any of them cried. Stankewitz was removed from his home at age six, after his mother gave him “a severe beating” with an electrical ironing cord.
That Stankewitz was severely emotionally damaged by his early childhood is well-supported by the record. A report from Stankewitz’s pre-first grade teacher stated that Stankewitz had “[m]any behavioral problems,” and would frequently engage in acts such as running out the door, yelling, kicking and screaming. According to a probation officer’s report, after Stankewitz was removed from his home he was taken to the pediatric ward, where the pediatric staff was unable to control him. Physical restraints had to be used and frequently replaced because Stankewitz repeatedly chewed through them. Stankewitz was removed from two foster homes for throwing chairs at and kicking his foster parents, running away and attacking probation offi*1169cers. Several other reports describe similar uncontrollable behavior after he was removed from his abusive home. Given the evidence in the record, the district court was clearly justified in crediting Stankewitz’s allegation that significant emotional damage followed from his troubled childhood. Indeed, contrary to the state’s argument, we have previously held that a district court commits error by “discount[ing] the significance of [a petitioner’s] early childhood on the ground that [he] has no recollection of the first few years of his life.” James v. Ryan, 679 F.3d 780, 815 (9th Cir.2012) (internal quotation marks omitted). “It is well established that early childhood trauma, even if it is not consciously remembered, may have catastrophic and permanent effects on those who ... survive it.” Id. (alter in original) (quoting Hamilton v. Ayers, 583 F.3d 1100,1132 (9th Cir.2009)).1
The state also challenges the district court’s findings that Stankewitz had a history of substance abuse and that Stankewitz consumed substantial quantities of alcohol, heroin and methamphetamine in the days leading to the murder. The state acknowledges that Stankewitz introduced evidence on remand that supports each of these allegations, but argues that some evidence in the record contradicts the district court’s findings. Even accepting the state’s argument that there is conflicting evidence in the record, the state has fallen well short of establishing clear error. See United States v. Working, 224 F.3d 1093, 1102 (9th Cir.2000) (en banc) (“Where there are two permissible views of the evidence, the factfinder’s choice between them cannot be clearly erroneous.” (quoting Anderson v. Bessemer City, 470 U.S. 564, 574, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985))).
IV.
The state also argues that the record, as developed on remand, does not establish deficiency or prejudice under Strickland. We address each of the state’s arguments in turn.
A. Deficiency
The state contends that Goodwin’s representation was not deficient for two reasons. First, it argues that his failure to investigate and present mitigating evidence was reasonable because, as Goodwin explained in a declaration, Stankewitz was opposed to any penalty phase defense. We considered and rejected this argument eight years ago. We wrote:
Goodwin’s acquiescence in Stankewitz’s purported opposition was not reasonable.
First, Stankewitz’s supposed opposition to “any penalty phase defense” is belied by the record. Goodwin did introduce penalty phase witnesses, including a member (by marriage) of Stankewitz’s family. Stankewitz, whose willingness to object verbally when he disagreed with the decisions of counsel *1170or the court was vividly demonstrated by the transcripts of both trials, did not object to this testimony. This suggests either that Stankewitz did not object to the testimony of family members or that Goodwin could have convinced Stankewitz to accept such evidence if Goodwin had conducted a proper investigation and presented the evidence to Stankewitz.
Second, Goodwin has alleged that Stankewitz did not want his family used as witnesses; but he does not claim that Stankewitz objected to his family being interviewed or to an investigation that relied on non-family members. We have previously held that opposition to calling family members or experts as witnesses does not excuse an attorney from interviewing experts and family members or from investigating documents containing mitigating evidence. See Silva v. Wood-ford, 279 F.3d 825, 840 (9th Cir.2002) (“Silva’s directive [against calling his family members as witnesses] did not automatically require foregoing all inquiry into his past.”); of Hayes v. Woodford, 301 F.3d 1054, 1067 (9th Cir.2002) (distinguishing Silva where defendant made clear to counsel that he did not want his family members called as witnesses or involved in any investigation). Stankewitz’s supposed opposition also should not have prevented Goodwin from attempting to rebut the prosecution’s aggravating evidence, such as by challenging Officer Reid’s testimony about the shoot-out, discussed above.
Thus, Stankewitz’s supposed opposition to mitigating evidence cannot explain Goodwin’s tactics.
Stankewitz, 365 F.3d at 721-22. The state did not introduce any evidence on remand nor has it advanced any argument on appeal that undermines our earlier analysis.2, 3 We thus reaffirm our determination that Stankewitz’s supposed opposition to a penalty phase defense does not excuse his failure to investigate and present mitigating evidence.
*1171Second, the state argues that “dramatic” changes to the record since we considered Stankewitz’s allegations now undermine our prior deficiency analysis. The critical change, according to the state, is that the record now establishes that Goodwin had in his possession Sciandra’s files from Stankewitz’s first trial that contained much of the mitigating evidence that Stankewitz now relies upon. According to the state, this establishes that Goodwin reasonably investigated mitigation evidence. We disagree. That Goodwin possessed some mitigating evidence does not alter our conclusion that he rendered deficient performance.
It is undisputed that, despite Goodwin’s possession of the files, he did not investigate any of the evidence contained within them. He did not contact Sciandra to discuss the contents of the files. He also never “hired an investigator or interviewed Stankewitz’s teachers, foster parents, psychiatrists, psychologists or anyone else who may have examined or spent significant time with him during his childhood and youth” and “did not interview anyone involved in Stankewitz’s first trial and thus did not know about the existence of any diagnoses of Stankewitz’s mental capabilities.” Id. at 719. He did not obtain a psychological examination of Stankewitz, despite his belief that Stankewitz was not mentally competent, and did not pursue any of the evidence of Stankewitz’s history of drug and alcohol abuse. See id.4 We thus reaffirm our holding that Goodwin’s failure to “take these steps to look into Stankewitz’s life history, despite tantalizing indications in the record, as in Wiggins [v. Smith, 539 U.S. 510, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003)], that ‘would lead a reasonable attorney to investigate further’ ” amounted to constitutional deficiency. Id. at 720 (quoting Wiggins, 539 U.S. at 527, 123 S.Ct. 2527); see also Lambright v. Schriro, 490 F.3d 1103, 1117 (9th Cir.2007) (reaffirming the principle that “when ‘tantalizing indications in the record’ suggest that certain mitigating evidence may be available, those leads must be pursued” (quoting Stankewitz, 365 F.3d at 720)).
The state’s argument that Goodwin’s mere possession of Sciandra’s files demonstrates that Goodwin conducted a reasonable investigation defies logic — if anything, that Goodwin had this evidence at his fingertips but did not investigate or present it is further proof of his deficiency. It also conflicts with our prior decisions. In James, 679 F.3d 780, for instance, the petitioner similarly argued that his counsel provided ineffective assistance during the penalty phase of his trial. Like Stankewitz, the petitioner proffered substantial mitigating evidence that could have been presented, which documented an impoverished childhood marred by alcoholic, criminal and abusive parents, signs of serious childhood trauma, placement in various foster homes and institutions, possible mental deficiencies, a long history of drug use and substantial drug use immediately prior to committing the crime of conviction. See id. at 810-15. In James, it was undisputed that the petitioner’s trial counsel possessed and was aware of much of the mitigating evidence that the petitioner proffered, including evidence that the petitioner suffered from mental illness, that he was under the influence of LSD at the time of the murder and that he suffered psychological and emotional trauma as a child. See id. at 808-09. We held that the petitioner received ineffective assistance because counsel’s possession of such evidence “should have prompted further in*1172vestigation.” Id. at 808. Counsel’s possession and awareness of the evidence, but failure to investigate or present it, is evidence of — not an excuse for — his deficiency-
In Summerlin v. Schriro, 427 F.3d 623 (9th Cir.2005) (en banc), the petitioner similarly argued that his trial counsel provided ineffective assistance by failing to investigate and present the same sort of classic mitigation evidence that we consider here. See id. at 631 (describing evidence of the petitioner’s “tortured family history, including the fact that [his] alcoholic mother beat him frequently,” and possible mental health issues). There, similar to here, one of the petitioner’s prior attorneys had investigated some mitigating evidence and communicated the results of the investigation to the petitioner’s trial counsel. See id. at 632. We held that the petitioner’s trial counsel rendered ineffective assistance because, despite the fact that the evidence “was in [his] hands,” he “failed to do any further investigation or development of this critical mitigation evidence.” Id.; see also Rompilla v. Beard, 545 U.S. 374, 385-90, 125 S.Ct. 2456, 162 L.Ed.2d 360 (2005) (holding that counsel rendered ineffective assistance where she obtained the petitioner’s prior conviction file but failed to adequately review it prior to the petitioner’s sentencing hearing); Correll v. Ryan, 539 F.3d 938, 944 (9th Cir.2008) (holding that trial counsel rendered ineffective assistance for failing to present mitigating evidence despite “the abundance of classic mitigation evidence of which [he] was aware”).
Finally, even if we were to assume that Goodwin was aware of all of the mitigating evidence in Sciandra’s files, he was still deficient for failing to present the evidence with no tactical basis for doing so. See Hamilton, 583 F.3d at 1113 (“Counsel ... has an obligation to present and explain to the jury all available mitigating evidence.” (citing Correll, 539 F.3d at 946)); id. at 1119 (“Defense counsel compounded the errors he committed during the investigative stage of the penalty phase by presenting almost none of the little mitigating evidence he had discovered.”).
We thus reject the state’s argument that Goodwin’s possession of Sciandra’s files undermines our prior analysis and hold that Goodwin provided deficient representation.5 In doing so, we remain sensitive to the Supreme Court’s guidance that habeas courts must give substantial weight to the constitutionally protected independence of counsel and the wide latitude counsel must have in making tactical decisions. See Cullen v. Pinholster, — U.S. -, 131 S.Ct. 1388, 1407, 179 L.Ed.2d 557 (2011). We are faced, however, with a situation in which counsel’s failure to investigate and present mitigating evidence cannot be rationalized on any tactical ground. It is simply untenable that Goodwin’s decision to forgo powerful mitigating evidence and instead put on his paltry penalty phase presentation was *1173made “in the exercise of reasonable professional judgment.” Id. at 1403.
B. Prejudice
The state’s final argument is that the Supreme Court’s decision in Wong v. Belmontes, 558 U.S. 15, 130 S.Ct. 383, 175 L.Ed.2d 328 (2009) (per curiam), undermines our prior prejudice analysis and the district court’s conclusion on remand that Stankewitz was prejudiced.6
In Belmontes the Supreme Court rejected a petitioner’s claim that his counsel was ineffective for failing to investigate and present additional mitigating evidence at the sentencing phase of his capital murder trial. See id. at 384. The Court assumed that Belmontes’ counsel’s failure to investigate satisfied Strickland’s deficiency prong, see id. at 386, but held that, even so, Belmontes could not show prejudice for two reasons. First, some of the evidence Belmontes advanced with his habeas petition was merely cumulative of the substantial humanizing evidence his counsel had already presented at trial. See id. at 387-88. The Court noted that Belmontes’ counsel presented “substantial” mitigating evidence: nine witnesses, including various family members, who testified about Belmontes’ terrible childhood, his alcoholic and abusive father, his strong relationships with certain family members, his religious conversion and his success at working as part of a firefighting crew. Id. at 387-88.
Second, the record made clear that presenting additional mitigating evidence would have opened the door to “potentially devastating” aggravating evidence that Belmontes was responsible for a second, unsolved murder — “the worst kind of bad evidence.” Id. at 385, 387-90. The evidence would have disclosed that Belmontes was suspected of killing a man “execution style,” that Belmontes possessed the gun used in the murder and that he had boasted to several people that he committed the murder. Id. at 385. Furthermore, Belmontes’ counsel specifically testified that he did not introduce additional mitigating evidence because he had “grave concerns” that presenting the evidence would open the door to the damaging evidence. Id.; see also id. at 386 (describing the trial court’s warnings to Belmontes’ counsel that he must tailor his mitigation case carefully to keep the evidence out). Considering all of these factors, the Supreme Court concluded that Belmontes suffered no prejudice, because “[i]t is hard to imagine expert testimony and additional facts about Belmontes’ difficult childhood outweighing the facts of [the second] murder.” Id. at 391.
The state contends that, applying Belmontes, Stankewitz likewise cannot establish prejudice. It argues that the evidence of Stankewitz’s history would have been damaging because the jury might have concluded from the evidence that Stankewitz had a violent, antisocial personality and it would have opened the door to further evidence of antisocial behavior.
We accept the state’s argument that some of the evidence Stankewitz has proffered illustrates serious antisocial behavior, including several emotional and violent outbursts throughout his life. We also accept the state’s argument that such evidence may be aggravating, rather than mitigating. See Daniels v. Woodford, 428 F.3d 1181, 1192-93, 1210 (9th Cir.2005) (suggesting that evidence that the defendant was a sociopath was aggravating); Beardslee v. Woodford, 358 F.3d 560, 583 (9th Cir.2004) (acknowledging that an antisocial personality diagnosis can be damaging); In re Crew, 52 Cal.4th 126, 127 Cal. *1174Rptr.3d 285, 254 P.3d 320, 333-34 (2011) (treating a defendant’s possible diagnosis with antisocial personality disorder as aggravating). But see Lambright, 490 F.3d at 1122, 1125 (treating antisocial or sociopathic behavior as a mitigating factor based on Arizona law).
Even so, Belmontes does not change our conclusion that Stankewitz was prejudiced by Goodwin’s failure to investigate or present any of the available mitigating evidence. Indeed, the contrast between the two cases reinforces our previous unanimous conclusion, when remanding the case to the district court, that a “more complete presentation, including even a fraction of the details Stankewitz now alleges, could have made a difference.” Stankewitz, 365 F.3d at 724.
Stankewitz’s case is materially different from Belmontes.7 First, unlike Belmontes, in which the mitigation presentation was substantial, here the mitigation presentation barely touched on Stankewitz’s extremely troubled childhood. See Stankewitz, 365 F.3d at 716-22. Second, in its penalty phase presentation, the prosecution put before the jury substantial evidence of Stankewitz’s violent, antisocial behavior. Multiple witnesses testified, for instance, that Stankewitz severely beat a 70-year-old man, stole his car and then participated in a highspeed chase and shootout that resulted in a police officer being shot in the head, with strong implications that Stankewitz was the shooter. See id. at 710-11,- 720, 723-24. Several other witnesses testified about Stankewitz’s attack on a youth counselor at the California Youth Authority, his armed robbery and kidnapping of Jesus Meraz, his stabbing of a fellow inmate, his attack on several police officers who were attempting to book him and various violent outbursts while he was in jail. See id. at 710-11, 723. To the extent additional evidence of the violent emotional outbursts that are part of Stankewitz’s history would have had an aggravating impact, it would have been marginal relative to the evidence of antisocial behavior already before the jury.
In short, Stankewitz’s posture at the penalty phase was the polar opposite of that in Belmontes. For Stankewitz, any adverse impact of the additional mitigation evidence would have been merely cumulative because the prosecution had already painted a grim picture of Stankewitz’s violent, antisocial tendencies. Instead, it is the mitigating effect of the proffered evidence that would have been novel because the jury had heard next to nothing about Stankewitz’s traumatic childhood. Furthermore, although the state has alluded to further aggravating evidence that it would have presented had Goodwin developed Stankewitz’s childhood history, it has not identified anything that would have been even remotely as damaging as the second murder counsel sought to keep out in Bel-, montes.8
*1175Accordingly, we continue to believe the more analogous cases here are the Supreme Court’s decisions in Wiggins v. Smith and Williams v. Taylor, 529 U.S. 362, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000). See Stankewitz, 365 F.3d at 714-16. In Wiggins, for instance, a capital habeas petitioner’s defense counsel failed to introduce social history mitigation evidence during the penalty phase, including evidence that “Wiggins experienced severe privation and abuse in the first six years of his life while in the custody of his alcoholic, absentee mother[, and that h]e suffered physical torment, sexual molestation, and repeated rape during his subsequent years in foster care.” Wiggins, 539 U.S. at 535, 123 S.Ct. 2527. The Court pointed out that this is the type of evidence that is “relevant to assessing a defendant’s moral culpability,” id., and held that the failure to introduce this evidence at the penalty phase was prejudicial: “[H]ad the jury been confronted with this considerable mitigating evidence, there is a reasonable probability that it would have returned with a different sentence.” Id. at 536, 123 S.Ct. 2527. So too, here. Stankewitz’s proffered mitigation evidence is precisely the sort that is relevant to his moral culpability. The Court did note that, because Wiggins did not have a pattern of aggressive behavior or a criminal history, his mitigation evidence was unlikely to pose the double-edged sword problem presented in other cases (and now exemplified by Belmontes). See id. at 515-16, 535-36, 123 S.Ct. 2527. Although Stankewitz’s history is certainly not benign like Wiggins’, the probability that the proffered mitigation evidence would have cut in the prosecution’s favor is low given that the jury was already aware of Stankewitz’s violent, antisocial behavior.
The decision facing Goodwin was whether to rebut the state’s substantial aggravating evidence by presenting a narrative that might have humanized Stankewitz to the jury or simply to ignore the state’s presentation. Goodwin defaulted to the latter, without any tactical basis for doing so. Cf. Summerlin, 427 F.3d at 635 (“The net result was that [the defendant] presented no affirmative evidence and no rebuttal evidence, although — as we have discussed — there was an abundance of available classic mitigation evidence concerning family history, abuse, physical impairments, and mental disorders.”).
Another indicator of prejudice, as we explained in our previous opinion, is the difficult time the jury had reaching a unanimous verdict on death. See Stankewitz, 365 F.3d at 724-25. Several jurors initially voted for life and one juror stated that “extensive deliberation” was required to convince a hold-out juror to vote for death. See id. at 725. “Had the jury been able to place [Stankewitz’s] excruciating life history on the mitigating side of the scale, there is a reasonable probability that at least one juror would have struck a different balance.” Wiggins, 539 U.S. at 537, 123 S.Ct. 2527.
Finally, the dissent argues that the district court did not apply the proper standard in its prejudice analysis because it did not consider the potential for the supposedly mitigating evidence to be perceived as aggravating evidence, and that we should remand for the court to do so. We agree that the proper legal standard requires consideration of both the potential aggravating impact and the potential mitigating impact of the proffered evidence. We disagree, however, that the district *1176court must perform another prejudice analysis.
Even assuming the district court applied the incorrect standard and erroneously neglected to consider the possibility that the proffered evidence would be perceived as aggravating, our analysis above illustrates why application of the correct standard would yield the same result. Cf. Agarwal v. Arthur G. McKee & Co., 644 F.2d 803, 807 & n. 3 (9th Cir.1981) (“Although the district court may have applied the wrong standard, application of the correct standard would surely have led to the same result.... Therefore it appears that under any standard the district court would reach the same result, and a remand under those conditions would be pointless.”).9
In sum, the record shows that there was indeed substantial mitigating evidence that could have been presented with little or no risk of further aggravating the negative information the jury already knew about Stankewitz. Given Goodwin’s paltry penalty phase presentation and the jury’s apparent difficulty in reaching a verdict, we hold that the district court correctly found that Goodwin’s failures prejudiced Stankewitz before the jury.
V.
Eight years ago, we recognized that Stankewitz advanced a colorable claim of ineffective assistance of counsel and we remanded for an evidentiary hearing to give the state an opportunity to rebut Stankewitz’s allegations. After agreeing to proceed without an evidentiary hearing and failing to meaningfully rebut Stankewitz’s allegations, the state asks us to remand for an evidentiary hearing so that it can try again. The state has given us no good reason to do so. We affirm the district court’s order granting Stankewitz a writ of habeas corpus directing the State of California to either: (a) vacate and set aside the death sentence in People v. Douglas Ray Stankewitz, Fresno County Superior Court Case No. 227015-5, unless the State of California initiates proceedings to retry Stankewitz’s sentence within 90 days; or (b) resentence Stankewitz to life without the possibility of parole.
AFFIRMED.
. Also, contrary to the state’s argument, the record contains significant evidence that Stankewitz's relationship with his troubled family did not end when he was removed from his home at age six. According to reports in the record, Stankewitz was returned to his mother’s custody for a short period of time at age 11. At age 12, Stankewitz was placed with his aunt, just before his mother was arrested and convicted of voluntary manslaughter for shooting and killing a man. At age 13, Stankewitz was released to his father, who had just been released from prison, but ran away after his father beat one of his brothers with a whip. At age 14, Stankewitz was with one of his brothers during a high speed chase, which culminated in a police officer being shot. Furthermore, evidence in the record indicates that Stankewitz continued to feel particularly attached to his ex-convict uncle and his brothers well into his teenage years.
. The state cites to Schriro v. Landrigan, 550 U.S. 465, 127 S.Cl. 1933, 167 L.Ed.2d 836 (2007), for the proposition that Goodwin’s failure to present additional mitigating evidence cannot be the basis for ineffective assistance under Strickland because Stankewitz expressed a desire not to present such evidence. We have already rejected this expansive reading of Landrigan, "a post-AEDPA case [in which] the defendant actively obstructed counsel’s investigation and outright refused to allow counsel to present any mitigating evidence.” Hamilton, 583 F.3d at 1119. As we noted, the defendant in Landrigan explicitly instructed witnesses not to testify and repeatedly interrupted his lawyer's presentation to the court. See id. We held that Landrigan is inapplicable where the defendant “did not threaten to obstruct the presentation of any mitigating evidence that counsel found.” Id. Here, the district court specifically found that "despite his alleged objection to the presentation of mitigation evidence, Stankewitz did not interrupt or try to sabotage trial counsel's presentation.” Landrigan is thus inapposite.
. For the first time since Stankewitz filed his petition, the state advances the argument that Goodwin made a "tactical” decision not to present any mitigating evidence because he knew Sciandra presented some of the mitigating evidence in Stankewitz’s first trial, in which Stankewitz nonetheless was sentenced to death. This argument, too, is belied by the record. Goodwin himself declared that he did not present the evidence because of Stankewitz’s opposition. The record also reveals that Goodwin was willing to present mitigating evidence insofar as it fell in his lap. See Stankewitz, 365 F.3d at 720-21 (describing how Stankewitz's sister-in-law became a witness because of a chance meeting with Goodwin). Furthermore, we reject the state’s suggestion that a lawyer exercises reasonable judgment when he or she decides not to present mitigating evidence simply because that mitigating evidence was unsuccessful in a pri- or trial.
. The only evidence that Goodwin looked at Sciandra’s files is a one-page chart he created, which lists the whereabouts of Stankewitz’s parents and siblings throughout the 1960s and 70s.
. The state points out that some of the allegations we considered in our prior decision were not credited by the district court on remand. In particular, the court considered questionable the declarations of Rosamond and Rosetta Bollmeyer, Stankewitz’s foster mother and sister, who testified that Stankewitz may have been sexually abused while he was institutionalized at Napa State Hospital. The district court also did not make specific findings with respect to specific facts that we mentioned, such as whether Stankewitz was taken to the emergency room three times before his first birthday, whether Stankewitz was born with fetal alcohol syndrome or whether a scar on Stankewitz’s head originated from child abuse. See Stankewitz, 365 F.3d at 717-18. The court’s failure to make a specific finding as to every one of Stankewitz’s allegations is not dispositive; we are satisfied that the findings the court did make are sufficient to sustain Stankewitz's ineffective assistance claim.
. Belmontes was decided between the time the district court issued its order granting Stankewitz’s petition for habeas corpus and the time it issued its order denying the state's motion for reconsideration.
. Our dissenting colleague argues that merely distinguishing Belmontes is insufficient because it did not mark "the exact boundary between prejudice and harmlessness.” Dissent at 1178. We agree there is no exact boundary, but we can certainly compare the circumstances the Court found dispositive in Belmontes with those that exist here and determine whether the cases are analogous. This is the very process we applied in our original opinion, looking to the relevant Supreme Court cases extant at the time and unanimously concluding not only that Stankewitz was prejudiced, but that "even a fraction of the details Stankewitz [previously] allege[d] could have made a difference.” Stankewitz, 365 F.3d at 724. Belmontes is a more recent comparator, but, as we explain, this case is nothing like Belmontes.
. The state argues that if Stankewitz had presented expert testimony that his emotional outbursts were the result of a mental disorder, it would have countered with expert testimony that the outbursts were the result of his antisocial personality. As discussed, there was already powerful evidence before the jury *1175of Stankewitz's antisocial behavior. The state also suggests that if Stankewitz had presented evidence of his troubled childhood to the jury, it would have attempted to impeach the credibility of some of the witnesses. This additional aggravating evidence is a far cry from evidence linking the petitioner to an additional murder, as in Belmontes.
. The dissent also urges remand because "appellate judges are ill-suited to apply the correct prejudice standard, in the first instance." Dissent at 1177. As discussed above, the district court did not conduct an evidentiary hearing, and the record is entirely documentary. Thus, a review of the record does not hinge on credibility determinations made of live witnesses. Appellate courts are competent to review documentary evidence, particularly in light of the thorough review of the record this same panel conducted when deciding the earlier appeal in this case. Furthermore, the district court did find, as a factual matter, that the state failed to rebut the majority of Stankewitz’s allegations — the same allegations that our panel unanimously held would have made a difference had they been presented to the jury during the penalty phase. See Stankewitz, 365 F.3d at 724.