Concurrence by
Judge BEA;Partial Concurrence and Partial Dissent by JudgeWARDLAW
ORDER
The opinion filed March 4, 2016, and reported at 815 F.3d 1233, is hereby amended concurrent with the filing of an Amended Opinion today. With these amendments, Judges Bea and N.R. Smith voted to deny the petition for rehearing en banc, and Judge Wardlaw voted to grant the petition for rehearing en banc.
The full court was advised of the petition for rehearing en banc. A judge requested a vote on whether to rehear the matter en banc. The matter failed to receive a majority of the votes of the nonrecused active judges in favor of en banc consideration. Fed. R. App. P. 35.
The petition for rehearing en banc is DENIED. No further petitions for rehearing or rehearing en banc may be filed in response to the amended disposition.
OPINION
N.R. SMITH, Circuit Judge:Petitioner Charles Michael Hedlund, an Arizona state prisoner, appeals the district court’s denial of his 28 U.S.C. § 2254 habe-as corpus petition. A jury convicted Hed-lund of one count of first degree murder for the 1991 killing of Jim McClain. The trial court sentenced Hedlund to death for the murder. The jury also convicted Hed-lund of the second degree murder of Christene Mertens.
The relevant state court decision, relating to Hedlund’s claims regarding (1) the use of a leg brace as a security measure during trial; (2) the use of dual juries; (3) juror bias; (4) counsel’s performance during the plea process; and (5) counsel’s performance during the penalty phase, was not contrary to, nor an unreasonable application of, clearly established federal law or based on an unreasonable determination of the facts before that court.1 See 28 U.S.C. § 2254(d).
*563However, the Arizona Supreme Court applied a “causal nexus” test, whereby not all mitigating evidence was considered under Lockett v. Ohio, 438 U.S. 586, 98 S.Ct. 2954, 57 L.Ed.2d 973 (1978), Eddings v. Oklahoma, 455 U.S. 104, 102 S.Ct. 869, 71 L.Ed.2d 1 (1982), and their progeny. See McKinney v. Ryan, 813 F.3d 798 (9th Cir.2015) (en banc). Therefore, such decision was contrary to clearly established federal law. See 28 U.S.C. § 2254(d). We must reverse the district court and remand with instructions to grant the petition with respect to Hedlund’s sentence.2
FACTS AND PROCEDURAL HISTORY
Findings of fact in the last reasoned state court decision are entitled to a presumption of correctness, rebuttable only by clear and convincing evidence. See Runningeagle v. Ryan, 686 F.3d 758, 763 n.1 (9th Cir. 2012); Moses v. Payne, 555 F.3d 742, 746 n.1 (9th Cir. 2008). Therefore, we adopt the statement of facts as presented by the Arizona Supreme Court in its 1996 opinion on consolidated direct appeal.
Beginning February 28, 1991, James Erin McKinney and Charles Michael Hedlund (Defendants) commenced a residential burglary spree for the purpose of obtaining cash or property. In the course of their extensive planning for these crimes, McKinney boasted that he would kill anyone who happened to be home during a burglary and Hedlund stated that anyone he found would be beaten in the head.
Defendants enlisted two friends to provide information on good burglary targets and to help with the burglaries. These two friends, Joe Lemon and Chris Morris, were not physically involved in the burglaries in which the murders occurred. It was from Lemon and Morris, however, that Defendants learned that Christene Mertens would make a good burglary target.
The first burglary in the spree occurred on February 28, 1991. Mertens’ home was the intended target that night, but she came home and scared the would-be burglars away. A different residence was chosen to burglarize, but Defendants obtained nothing of value. Both Defendants, as well as Lemon and Morris, were involved in this crime.
The second and third burglaries occurred the next night, March 1. This time Lemon was not involved. The three participants stole a .22 revolver, $12, some wheat pennies, a tool belt, and a Rolex watch.
A. The first murder
The fourth burglary took place on March 9, 1991. This time only McKinney and Hedlund were involved. Mertens was picked again because Defendants had been told by Lemon and Morris, who knew Mertens’ son, that Mertens kept several thousand dollars in an orange juice container in her refrigerator.
Mertens was home alone when Defendants entered the residence and attacked her. Beaten and savagely *564stabbed, Mertens struggled to save her own life. Ultimately, McKinney held her face down on the floor and shot her in the back of the head, covering his pistol with a pillow to muffle the shot. Defendants then ransacked the house and ultimately stole $120 in cash.
B. The second murder
Defendants committed the fifth burglary on March 22, 1991. The target was Jim McClain, a sixty-five-year-old retiree who restored cars for a hobby. McClain was targeted because Hedlund had bought a car from him some months earlier and thought McClain had money at his house. Entry was gained through an open window late at night while McClain was sleeping. Hedlund brought along his .22 rifle, which he had sawed-off to facilitate concealment. Defendants ransacked the front part of the house then moved to the bedroom. While he was sleeping, McClain was shot in the back of the head with Hedlund’s rifle. Defendants then ransacked the bedroom, taking a pocket watch and three hand guns; they also stole McClain’s car.
State v. McKinney, 185 Ariz. 567, 917 P.2d 1214, 1218-19 (1996) (en banc), superseded by statute on other grounds as stated in State v. Martinez, 196 Ariz. 451, 999 P.2d 795, 806 (2000) (en banc).
Hedlund and McKinney were each indicted on two counts of first degree murder and four other counts relating to the robberies. Both Defendants were tried in the same courtroom before dual juries. Before returning its verdict, Hedlund’s jury asked whether he could “be convicted as an accomplice to the burglary and not be convicted in the murder charge.” On November 12, 1992, the jury found Hed-lund guilty of the second-degree murder of Mertens, the first-degree murder of McClain, and lesser charges. In a special verdict, the jury unanimously found that Hedlund was guilty of the premeditated murder of McClain, rejecting a felony murder theory. The trial court sentenced Hedlund to death for the first degree murder of McClain and to terms of imprisonment on the lesser charges.
Upon direct appeal, the Arizona Supreme Court affirmed the conviction and sentence. McKinney, 917 P.2d at 1214. In its opinion, the Arizona Supreme Court considered five claims relevant to this appeal: (1) whether the use óf dual juries deprived Hedlund of his right to a fair trial, (2) whether ordering Hedlund to wear a visible leg restraint during trial deprived Hedlund of his right to a fair trial, (3) whether Hedlund was denied his right to a fair and impartial jury when the trial court refused to dismiss a juror distantly related to one of the victims, (4) claims surrounding the negotiation of a second plea deal, and (5) the consideration and weighing of aggravating and mitigating factors.
The Arizona Supreme Court denied relief on all claims and noted “ample evidence” that Hedlund killed McClain, including: Hedlund’s finger and palm prints were on McClain’s briefcase, which had been rifled during the burglary; Hedlund’s fingerprints were on the magazine of his sawed-off rifle; the bullet that killed McClain was consistent with having come from Hedlund’s rifle; Hedlund had modified his rifle by sawing it off in order to conceal it; Hedlund hid the rifle after the murder; Hedlund asked Morris to get rid of the rifle before - police found it; and Hedlund expressed remorse after he was arrested.
After the Arizona Supreme Court rejected Hedlund’s claims, Hedlund filed a petition for post-conviction relief (PCR) and then an amended PCR petition in the state trial court. On PCR review, the trial court denied the amended petition without an *565evidentiary hearing. The Arizona Supreme Court summarily denied Hedlund’s petition for review.
On August 5, 2003, Hedlund filed the operative amended petition for a writ of habeas corpus in federal district court. Hedlund later filed a motion to expand the record and for evidentiary development as to certain claims. On March 31, 2005, the district court denied the motion to expand the record and denied six of Hedlund’s claims. On August 10, 2009, the district court denied Hedlund’s remaining claims, found Hedlund not entitled to habeas relief, and entered judgment.
The district court granted a certificate of appealability (COA) on three claims. We expand the COA to include three additional claims, as explained below. We otherwise deny Hedlund’s request to expand the COA.
STANDARD OF REVIEW
“We review de novo the district court’s decision to grant or deny a petition for. writ of habeas corpus.” Rhoades v. Henry, 598 F.3d 495, 500 (9th Cir. 2010). Because Hedlund initiated district court proceedings in 2002, the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA) applies. See Lindh v. Murphy, 521 U.S. 320, 336-37, 117 S.Ct. 2059, 138 L.Ed.2d 481 (1997). “Before we can apply AEDPA’s standards, we must identify the state court decision that is appropriate for our review.” Barker v. Fleming, 423 F.3d 1085, 1091 (9th Cir. 2005). Although “AEDPA generally requires federal courts to review one state decision,” where “the last reasoned decision adopted or substantially incorporated the reasoning from a previous decision ... it [is] reasonable for the reviewing court to look at both decisions to fully ascertain the reasoning of the last decision.” Id. at 1093; see also Amado v. Gonzalez, 758 F.3d 1119, 1130 (9th Cir. 2014) (“When the last reasoned decision is a state appellate court decision which ‘adoptfe]’ or ‘substantially incorporate^]’ lower state court decisions, we may review those lower state court decisions as part of our review of the state appellate court’s decision.” (quoting Barker, 423 F.3d at 1093)); Lewis v. Lewis, 321 F.3d 824, 829 (9th Cir. 2003) (Although this court was required to review the last reasoned opinion of the state appellate court, our analysis “necessarily include[d] discussion of the trial court’s decision” “[bjecause th[e appellate] decision affirmed the trial court and adopted one of the reasons cited by the trial court.”).
A petitioner must overcome a high threshold to obtain relief under AEDPA:
Federal habeas relief may not be granted for claims subject to [28 U.S.C.] § 2254(d) unless it is shown that the earlier state court’s decision was contrary to federal law then clearly established in the holdings of [the Supreme] Court, § 2254(d)(1); or that it involved an unreasonable application of such law,' § 2254(d)(1); or that it was based on an unreasonable determination of the facts in light of the record before the state court, § 2254(d)(2).
Harrington v. Richter, 562 U.S. 86, 100, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011) (internal quotation marks and citation omitted). “[T]he only definitive source of clearly established federal law under AEDPA is the holdings (as opposed to the dicta) of the Supreme Court as of the time of the state court decision.” Clark v. Murphy, 331 F.3d 1062, 1069 (9th Cir. 2003), overruled on other grounds by Lockyer v. Andrade, 538 U.S. 63, 71, 123 S.Ct. 1166, 155 L.Ed.2d 144 (2003).
If Supreme Court “cases give no clear answer to the question presented, ... it cannot be said that the state court *566unreasonably applied clearly established Federal law.” Wright v. Van Patten, 552 U.S. 120, 126, 128 S.Ct. 743, 169 L.Ed.2d 583 (2008) (internal quotation marks and alterations omitted). In other words, “[i]t is not an unreasonable application of clearly established Federal law for a state court to decline to apply a specific legal rule that has not been squarely established by [the Supreme Court].” Harrington, 562 U.S. at 101, 131 S.Ct. 770 (alterations omitted) (quoting Knowles v. Mirzayance, 556 U.S. 111, 122, 129 S.Ct. 1411, 173 L.Ed.2d 251 (2009)).
In cases where a petitioner identifies clearly established federal law and challenges the state court’s application of that law, our task under AEDPA is not to decide whether a state court decision applied the law correctly. See id. Rather, we must decide whether the state court decision applied the law reasonably. See id. (“[A]n unreasonable application of federal law is different from an incorrect application of federal law.” (quoting Williams v. Taylor, 529 U.S. 362, 410, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000))). If the state court applied the law reasonably, we must deny relief. See id. Thus, relief is proper only “in cases where there is no possibility fair-minded jurists could disagree that the state court’s decision conflicts with [the Supreme Court’s] precedents.” Id. at 102, 131 S.Ct. 770.
DISCUSSION
I. Visible Leg Brace at Trial
A. Background and procedural history
The trial court ordered both Hedlund and McKinney to wear a leg brace during trial, because it was important to courtroom security. During a pretrial hearing, Deputy Sheriff Jack Roger Lane testified that he was aware of a 1992 escape plot by Hedlund and McKinney. The plan was to “jump one of the guards, take his uniform and his weapon and one of them would put the-uniform on and they would walk out together. They would handcuff the guard and leave him there.” Lane received this information thirdhand from a subordinate officer, who heard it from an inmate. McKinney was specifically identified in the plot. The other individual was someone “charged with murder,” but Hedlund was not specifically named in the discussion on the record.3 Although Lane could not confirm it, the prosecutor was aware of an earlier escape attempt by McKinney during the summer of 1991.
Hedlund’s counsel challenged the leg brace, arguing that McKinney was the flight risk, not Hedlund. Recognizing its responsibility to maintain courtroom security, the trial court found it would be “irresponsible” to ignore the nature of the charges filed and the fact that both Defendants would be in close proximity to the jurors, staff, and others. The court denied the request to remove Hedlund’s leg brace, finding “reasonably reliable evidence that there is indeed a real escape risk in this case.” The court concluded that the leg brace was “a reasonable alternative to any other type of restraint that could be imposed on [Hedlund and McKinney] to assist in the preservation of a safe environment for everyone [in ■ the courtroom].” *567The court also attempted to minimize any potential prejudice by making the leg brace less visible. The court ordered new defense tables with backs covering two feet of the four-foot gap between the table top and the floor. The court also ensured that the Defendants would be seated in the courtroom before the juries arrived so the jurors would not see the Defendants walking stiff-legged in the braces.4
Hedlund’s counsel later filed multiple written motions objecting to the leg brace. During a post-trial evidentiary hearing, the court called Officer Richard Morris, one of the deputies present during trial. Officer Morris testified that during trial he was able to see the leg brace, similar to what was shown in a picture taken from the jury box. Hedlund’s investigator testified that she spoke with several jurors regarding the leg brace. The jurors agreed that it was understandable that the Defendants (who had been charged with such serious crimes) were put in some sort of restraint. While the restraints seemed to provide a sense of security to the jurors, the jurors stated that the leg brace did not have any impact on their verdict.
On Hedlund’s motion for new trial, after considering the escape risk by two Defendants charged with serious crimes and considering all of the various options (including limiting or increasing the number of deputies in the courtroom), the court concluded that the leg braces were proper to ensure the safety of the jurors, court staff, and everyone in the courtroom. While Hedlund could have helped facilitate concealment of the leg brace, the court noted that the leg brace did not “overwhelm” the jury to cause them to convict Hedlund on all charges.
On direct appeal, the Arizona Supreme Court credited the trial court’s record of security concerns, noting that “Hedlund attempted an escape during the summer of 1991 and also made plans with another capital defendant to escape by attacking a guard and taking his uniform and gun.”5 The court concluded that the leg restraint was not an abuse of discretion, given the trial judge’s well-founded security concerns and the absence of specific prejudice to Hedlund.
On habeas review, the federal district court noted that the Arizona Supreme Court erroneously attributed the 1991 escape attempt to Hedlund. However, the district court found no indication, let alone clear and convincing evidence, that the state court erred in finding both Hedlund and McKinney involved in the 1992 escape plot.
B., Hedlund’s leg restraint was not imposed based on a clearly unreasonable determination of the facts, nor was its imposition contrary to, or an unreasonable application of, clearly established federal law.
1. Standard of Review
As an initial matter, Hedlund argues that we should review this claim de novo because the Arizona Supreme Court erroneously attributed McKinney’s 1991 escape attempt to Hedlund. While the Arizona Supreme Court’s recitation of that *568fact is in error, as the federal district court correctly recognized, there is no indication that the trial court or the Arizona Supreme Court on direct review erred in concluding that Hedlund was involved in the 1992 escape plot with McKinney. The trial court presumed-that Hedlund was the other capital inmate plotting an escape with McKinney in 1992. Hedlund has not shown that this presumption was an unreasonable determination of the facts. Nor has he rebutted this factual determination with clear and convincing evidence.
Deputy Lane testified that an inmate (who knew McKinney) overheard McKinney plotting with another capital defendant. While the inmate-informant did not know Hedlund by name, jail security personnel drew the inference that the unnamed-capital defendant was Hedlund. Jail security personnel then acted upon this tip by noting the security risk on Hedlund’s jail card. Thus, when the Arizona Supreme Court stated that Hedlund made plans with another capital defendant (i.e., McKinney) to escape, this was neither factually erroneous nor objectively unreasonable based on Deputy Lane’s testimony.6
2. An essential state interest justified the leg restraint.
The Arizona Supreme Court’s decision affirming the use of the leg brace was not contrary to, or an unreasonable application of, clearly established federal law. Relevant to the state interest in requiring Hedlund to wear the leg brace, the Arizona Supreme Court recognized the need to leave courtroom security matters to the trial court’s discretion and upheld the trial court’s decision based on the fact that “the trial judge specifically made a record to document ... [his] well-founded security concerns.” Because the state appellate court affirmed based on the trial court’s security concerns, our review of the appellate court’s reasoned decision necessarily considers the trial record reflecting those concerns. Ordering the leg brace was justified by an essential state interest. The Supreme Court has defined shackling as “the sort of inherently prejudicial practice that ... should be permitted only where justified by an essential state interest specific to each trial.”7 Holbrook v. Flynn, 475 U.S. 560, 568-69, 106 S.Ct. 1340, 89 L.Ed.2d 525 (1986) (emphasis added). This determination turns on the facts of the case. Where an obstreperous defendant’s actions threaten the proceedings, even fully binding and gagging the defendant could be constitutionally permissible. Illinois v. Allen, 397 U.S. 337, 344, 90 S.Ct. 1057, 25 L.Ed.2d 353 (1970).
Here, the trial court found that Hedlund posed a security risk, thus warranting the *569minimally intrusive restraint. The trial court based this finding on the alleged 1992 escape plot involving both Defendants, the nature of the charges, and the safety of all persons in the courtroom during trial. The trial court’s conclusion, that specific security interests presented by the facts of this case warranted the leg restraint, was not contrary to, or an unreasonable application of, Holbrook (i.e., whether an essential state interest justified the use of a leg brace in this case). Holbrook, 475 U.S. at 569, 106 S.Ct. 1340; see also Hamilton v. Vasquez, 882 F.2d 1469, 1471 (9th Cir. 1989) (“Shackling is proper where there is a serious threat of escape or danger to those in and around the courtroom, or where disruption in the courtroom is likely if the defendant is not restrained.”); Crittenden v. Ayers, 624 F.3d 943, 971 (9th Cir. 2010) (“[Defendant] fail[ed] to rebut by clear and convincing evidence the trial court’s finding on the record that the restraints were justified by a state interest specific to Crittenden’s trial, namely his likelihood of escape or ‘nonconforming conduct.’ ”).
The record shows that jail personnel became aware of the 1992 escape plan after a tip from another inmate. While the inmate knew McKinney’s name, the inmate knew only that the co-plotter was another inmate charged with capital murder. Jail personnel then reviewed and acted upon this information. We do not know how jail personnel made the inference that the second inmate was Hedlund (e.g., whether Hedlund was the only other capital murder defendant who had been talking to McKinney, or was the only capital murder defendant housed in close proximity to McKinney). However, we do know that, after learning of the plot, jail personnel applied special security procedures to both Defendants and provided this information to the trial court.
While the trial court based its conclusion regarding the escape plot on information provided by jail personnel, the trial court’s reliance on this testimony was not contrary to, or an unreasonable application of, clearly established federal law. The trial court could have used the jail’s security-based decision as support for its conclusion that Hedlund posed an escape risk, because such decisions are subjective and discretionary. Cf. Rhodes v. Chapman, 452 U.S. 337, 349 n.14, 101 S.Ct. 2392, 69 L.Ed.2d 59 (1981) (“[A] prison’s internal security is peculiarly a matter normally left to the discretion of prison administrators.”).
The trial court relied on Deputy Lane’s assertion and concluded as follows:
I have been provided with what I have weighed and considered as reasonably reliable evidence that there is indeed a real escape risk in this case; perhaps not in the courtroom, but one that has been articulated outside the hearing of the Court in a fashion that indicates that both defendants were anticipated to be involved in it.... [There was] certainly some thought being given on the nature and mode of escape.
Although the trial court based this decision on hearsay coming from within the jail, there is no clearly established federal law suggesting that such' a finding is impermissible. Challenging the, trial court’s reliance upon such hearsay, Hedlund cites Gonzalez v. Pliler, 341 F.3d 897, 902 (9th Cir. 2003). However, Gonzalez is inapplicable to this case. First, Gonzalez applies the “less restrictive alternatives” test that was not clearly established federal law for AEDPA purposes. See Crittenden, 624 F.3d at 971-72 (recognizing that “case law requiring a court to weigh the benefits and burdens of shackling and pursue less restrictive alternatives was not clearly established federal law” before Deck v. Mis*570souri, 544 U.S. 622, 125 S.Ct. 2007, 161 L.Ed.2d 953 (2005)). Second, while Gonzalez recognized that the rules regarding physical restraints in California and the Ninth Circuit are largely coextensive, 341 F.3d at 901 n.1, the language stating that a court may not rely upon “the unsubstantiated comments of others” is drawn from California precedent, not clearly established federal law, id. at 902 (quoting People v. Mar, 28 Cal.4th 1201, 124 Cal.Rptr.2d 161, 52 P.3d 95, 107 (2002)).
It was not objectively unreasonable for the Arizona Supreme Court to find an essential state interest based on Lane’s testimony regarding the 1992 Hed-lund/McKinney escape attempt. Therefore, upholding the decision to impose the leg brace was not contrary to, or an unreasonable application of, clearly established federal law.
3. Prejudice
Because the Arizona Supreme Court’s adoption of the finding that Hedlund’s leg brace was justified by an essential state interest is not contrary to, or an unreasonable application of, Holbrook, we do not reach the question of prejudice.
II. Use of Dual Juries
A. Background and procedural history
Over the Defendants’ and prosecutor’s objections, the trial court ordered the Defendants’ cases tried before dual juries. The trial court reasoned that two trials would cause needless duplication, the victims’ families would suffer twice, and the only evidence that was not admissible to both juries could be covered in a single afternoon.8 The court set forth detailed procedures to be used at trial to avoid any problems.9
Hedlund challenged the use of dual juries in a special action to the Arizona Court of Appeals. See Hedlund v. Sheldon, 173 Ariz. 143, 840 P.2d 1008, 1009 (1992) (en banc). The Court of Appeals reversed, holding that the trial court exceeded its authority under the Arizona Rules of Criminal Procedure and the Arizona Supreme Court’s decision in State v. Lam-bright. Id. However, the Arizona Supreme Court reversed the Court of Appeals,10 *571concluding that the decision to empanel a dual jury is an “exercise of an individual judge’s discretion to use a particular technique in order to meet a specific problem in a single case.” Id. at 1011 (internal quotation marks omitted). Thus, the court affirmed the decision to empanel dual juries.
Post-trial, the trial court rejected Hed-lund’s renewed dual jury challenge. The court found that it had eliminated the risk of possible prejudice by empaneling dual juries rather than having one jury consider both Defendants’ guilt. The court concluded that this strategy worked, because the verdicts reflected that the juries were able to do their jobs intelligently.
B. The use of dual juries at trial was not contrary to, or an unreasonable application of, clearly established federal law.
Because Hedlund cannot point to clearly established federal law governing this claim, habeas relief is unavailable. The Supreme Court has not spoken on the issue of dual juries, and Hedlund cites no relevant authority.
In Zafiro v. United States, 506 U.S. 534, 538-39, 113 S.Ct. 933, 122 L.Ed.2d 317 (1993), the Court held that severance is not required in the face of antagonistic defenses. Even where prejudice is shown, Rule 14 of the Federal Rules of Criminal Procedure “leaves the tailoring of the relief to be granted, if any, to the district court’s sound discretion.” Id. at 539, 113 S.Ct. 933. The Court went on to say that severance should be granted “only if there is a serious risk that a joint trial would compromise a specific trial right of one of the defendants, or prevent the jury from making a reliable judgment about guilt or innocence.” Id.
Hedlund argues that this claim is valid, because certain of his specific trial rights were violated. While Zafiro holds that severance should be granted if there were a serious risk that a specific trial right would be compromised, Zafiro does not apply to § 2254 cases. Zafiro was a direct appeal originating in federal district court (i.e., a case in which the Federal Rules of Criminal Procedure applied). See Collins v. Runnels, 603 F.3d 1127, 1131-32 (9th Cir. 2010) (“By its own wording, Zafiro only applies to federal and not state court trials. It analyzes only the Federal Rules of Criminal Procedure applicable to federal district courts.”).
Even if we could apply Zafiro’s prejudice holding, Hedlund has not identified any specific constitutional right that has been violated. While he alludes to several constitutional violations, none of these arguments is well developed with citation to authority.. To the extent Hedlund argues that the prosecutor was improperly allowed to ask leading questions or elicit ambiguous testimony, he does not cite specific examples. Moreover, defense counsel had the opportunity to object at trial and did so. Although some objections were overruled, it is not clear the subject questions were leading or ambiguous, and if so, whether these evidentiary rulings were improper or harmed Hedlund in any way.
Even if ambiguous testimony or leading questions could somehow amount to a constitutional violation, the testimony did not prevent Hedlund from demonstrating lack of motive or putting on a full defense. The jury heard testimony that Hedlund' had a steady job and did not need to steal for money, and Lemon and Morris testified that Hedlund wanted nothing to do with the early burglaries.
Hedlund’s antagonistic defenses argument similarly fails. There is no constitutional right to severance merely because codefendants point the finger at each oth*572er. Moreover, the trial court’s remedy of employing procedural safeguards for the use of dual juries was within its discretion. Because none of Hedlund’s dual jury arguments demonstrate prejudice that is so “clear, manifest or undue that he was denied a fair trial,” even if Zafiro applied, this claim fails. See Lambright v. Stewart, 191 F.3d 1181, 1185-87 (9th Cir. 1999) (dual juries are permissible in capital cases so long as they comport with due process; denial of a motion to sever for antagonistic defenses not reversible without a showing of clear prejudice).
III. Juror Bias
A. Background and procedural history
On the second day of trial, one juror (“the Juror”) wrote a letter to the trial court disclosing the fact that she discovered she was distantly related to McClain, the second murder victim. In the letter, the Juror explained that she had become aware of this fact only that morning. When the Juror informed her mother she was serving on a jury, her mother stated that “she.had read of a trial starting in Mesa in which one of the victims had been married to a cousin of [the Juror’s] stepfather.” The Juror told her mother she could not discuss the trial and did not want to hear anything further. However, the Juror realized she would have to disclose this to the judge, so she asked her mother the name of the victim who was married to the stepfather’s cousin. The Juror stated that she didn’t personally recognize the name of the victim and had “never met, nor even heard of, [her] stepfather’s cousin, who is deceased.” She then concluded with the following statement regarding her ability to serve on the jury: “I don’t believe it would affect my ability to be fair and impartial, but I do not wish to compromise the proceedings in any way, so I wish to make the court aware of the situation.”
In response to the letter, the trial court held a hearing in chambers to explore whether the Juror should remain on the jury. The court read the Juror’s statement about impartiality back to her and asked if this was her belief. She responded, ‘Tes, it is.” In response to the court’s questions, the Juror explained that she had never met her stepfather’s now-deceased cousin who used to be married to McClain. In fact, until the conversation with her mother, she didn’t even know the cousin existed. Hedlund’s counsel inquired about the Juror’s relationship with her stepfather. The Juror explained that they “have a very superficial relationship.”
Hedlund’s counsel moved to strike the Juror for cause on the basis that she was a distant relative of the victim. The court stated, “given what she said here today I would not, based on what I’ve heard ... have stricken her for cause.... She is now on the jury. And based on the circumstances she has relayed to me, I’m going to deny the motion. She’ll remain on the panel.”
On appeal, the Arizona Supreme Court affirmed, finding that nothing in the record suggested the Juror was untruthful in stating she could be fair and impartial. The federal district court agreed. The district court found no risk of “substantial emotional involvement based on [the Juror’s] highly attenuated connection with the victim, about which the [J]uror was not even aware.”
B. The trial court complied with clearly established federal law when it determined no juror bias was present.
1. Hedlund has failed to prove actual bias.
Because the trial court followed clearly established federal law regarding actual *573juror bias, Hedlund’s claim fails. In Remmer v. United States, the Supreme Court held that juror bias should be determined “in a hearing with all interested parties permitted to participate.” 347 U.S. 227, 230, 74 S.Ct. 450, 98 L.Ed. 654 (1954). In Smith v. Phillips, the Supreme Court reversed a grant of habeas where the lower federal courts found insufficient a hearing to determine juror bias. 455 U.S. 209, 214-16, 221, 102 S.Ct. 940, 71 L.Ed.2d 78 (1982). During the Smith trial, one of the jurors applied for a job as an investigator with the district attorney’s office. Id. at 212, 102 S.Ct. 940. The prosecutors were aware of the application, but did not tell the court or defense counsel until after the jury returned its verdict. Id. at 212-13, 102 S.Ct. 940. Upon learning of the juror’s job application, the defendant moved to set aside the verdict. Id. at 213, 102 S.Ct. 940. The trial court held a hearing on this motion, at which both the prosecutors and the juror testified. Id. After the hearing, the trial court found that the juror was not biased as a result of his job application to the district attorney; and no evidence suggested a “sinister or dishonest motive” on the prosecutors’ part. Id. at 214, 102 S.Ct. 940. On habeas review, the federal district court found the trial court’s bias hearing insufficient and granted relief, which the Second Circuit affirmed.
The Supreme Court reversed the lower federal courts, finding that the trial court’s hearing (exploring the issue of juror bias) was sufficient to comply with due process. Id. at 221, 102 S.Ct. 940. The Court reiterated that it “has long held that the remedy for allegations of juror partiality is a hearing in which the defendant has the opportunity to prove actual bias.” Id. at 215, 102 S.Ct. 940. The Court rejected the argument that a trial court “cannot possibly ascertain the impartiality of a juror by relying solely upon the testimony of the juror in question.” Id. The Court disagreed that “the law must impute bias to jurors” in this situation. Id. Rather than ordering a new trial any time the issue of juror bias arises, the Court explained that holding a hearing to determine actual bias, such as that conducted by the trial court, is the appropriate course of action. Id. at 217, 102 S.Ct. 940.
The Smith Court concluded:
[D]ue process does not require a new trial every time a juror has been placed in a potentially compromising situation. Were that the rule, few trials would be constitutionally acceptable. The safeguards of juror impartiality, such as voir dire and protective instructions from the trial judge, are not infallible; it is virtually impossible to shield jurors from every contact or influence that might theoretically affect their vote. Due process means a jury capable and willing to decide the case solely on the evidence before it, and a trial judge ever watchful to prevent prejudicial occurrences and to determine the effect of such occurrences when they happen. Such determinations may properly be made at a hearing like that ordered in Remmer and held in this case.
Id. (footnote omitted).
The Court recognized that hearings of this sort will “frequently turn upon testimony of the juror in question,” but rejected the contention that “such evidence is inherently suspect.” Id. at 217 n.7, 102 S.Ct. 940. When a juror tries “as an honest man to live up to the sanctity of his oath[, the juror] is well qualified to say whether he has an unbiased mind in a certain matter.” Id. Lastly, the Court reiterated that, because the case was a § 2254 proceeding, the trial judge’s findings were “presumptively correct” and could not be overcome without clear and convincing evidence. Id. at 218, 102 S.Ct. 940.
*574The Arizona Supreme Court’s finding that the trial court did not abuse its discretion in refusing to dismiss the Juror was not contrary to, nor an unreasonable application of, Smith and Remmer. The trial judge conducted a hearing involving all interested parties to explore the issue of juror bias. At this hearing, Hed-lund had the opportunity to prove actual bias. This is the remedy prescribed by the Supreme Court. Id. at 215, 102 S.Ct. 940.
Hedlund challenges the sufficiency of the in-chambers hearing, arguing that the hearing was cursory, defense counsel was not given time to prepare, and it was the judge’s duty to question the Juror sufficiently. Hedlund argues that defense counsel could not be expected to conduct a vigorous cross-examination that might place Hedlund in a negative light. However, Smith does not dictate that an in-chambers hearing is insufficient, must be of a particular length, or must be conducted only after certain notice. Id.; see also Dyer v. Calderon, 151 F.3d 970, 974-75 (9th Cir. 1998) (“An informal in camera hearing may be adequate for this purpose; due process requires only that all parties be represented, and that the investigation be reasonably calculated to resolve the doubts raised about the juror’s impartiality.”). Here, the trial court questioned the Juror about her ability to be impartial, it did not rush defense counsel as counsel familiarized himself with the Juror’s letter, and it followed up with additional questions. Based on the Juror’s responses that she was unaware of both her stepfather’s now-deceased cousin and the victim, her relationship with her stepfather was superficial, and her belief was that she could remain impartial, the court was satisfied that no actual bias was present. As we explained in Calderon: “So long as the fact-finding process is objective and reasonably explores the issues presented, the state trial judge’s findings based on that investigation are entitled to a presumption of correctness.” 151 F.3d at 975. Thus, the court complied with clearly established federal law.
Although the Juror stated that she “believed” she could be impartial, she did not equivocate and the judge found this affirmation sufficient. Hedlund points to no authority requiring more of an assurance from the Juror. See Bashor v. Risley, 730 F.2d 1228, 1237 (9th Cir. 1984) (no error in keeping juror when juror responded to the question whether she could be impartial with, ‘Tes, I think I could.”).11
*5752. There is no clearly established law governing implied bias, and Hed-lund has not shown that implied bias should apply here.
There is no clearly established federal law regarding the issue of implied bias. The Supreme Court has never explicitly adopted or rejected the doctrine of implied bias. See Fields v. Woodford, 309 F.3d 1095, 1104 (9th Cir.) (noting that the “Supreme Court has never explicitly adopted (or rejected) the doctrine of implied bias”), amended by 315 F.3d 1062 (9th Cir. 2002). Thus, Hedlund’s claim fails on grounds of implied bias.12
Although we have presumed bias on a rare occasion, we have based this finding on close relationships or the fact that a juror has lied. See, e.g., United States v. Allsup, 566 F.2d 68, 71-72 (9th Cir. 1977) (bias of bank teller employees presumed where defendant robbed another branch of same bank and tellers had “reasonable apprehension of violence by bank robbers”); Green v. White, 232 F.3d 671, 676-78 (9th Cir. 2000) (presuming bias biased on juror’s pattern of lies). However, these cases are not clearly established federal law. In any event, nothing in the record suggests the Juror lied during voir dire or had a close relationship with McClain.
IV. Ineffective Assistance of Counsel During Plea Process13
A. Background and procedural history
Before trial, Hedlund reached a plea deal with the prosecutor. During an informal chambers discussion, defense counsel and the prosecutor were asked to explain the factual basis for the plea, which offered a guilty plea for the second degree murder of Mertens and theft with a prior for taking McClain’s guns. The trial court rejected the plea agreement, because it did not involve enough accountability for the McClain homicide. The court suggested a plea involving a burglary count with respect to McClain could be considered. However, as discussed below, the court had other reservations with respect to this and any future plea agreement. The parties continued negotiating and reportedly arrived at a second agreement consisting of a guilty plea for the second degree murder of Mertens, and theft with a prior and burglary non-dangerous with respect to McClain.
*576On the day the second plea was to be presented in chambers, Hedlund’s counsel instead called chambers and asked the judge if he would recuse himself. When the judge responded that he would not, Hed-lund filed a motion for recusal of judge, followed by a motion for change of judge. A second judge heard the latter motion. The motion made clear that Hedlund wanted to plead guilty to the new plea agreement, but that he refused to do so in front of the trial judge, Judge Sheldon. The second judge denied the motion and trial began immediately. The substance of the motion hearing is discussed below in the context of the ineffective assistance of counsel analysis.
On appeal, the Arizona Supreme Court questioned whether a second plea was ever reached. The court also noted that the prosecutor’s testimony at the hearing on the change-of-judge motion was that Hed-lund in fact rejected the second plea. Thus, the court rejected the claim that the trial court erred in any way with respect to the purported second plea. The claim challenging counsel’s performance was similarly rejected on PCR review.
B. The state PCR court did not unreasonably apply Strickland.
The two-part test for demonstrating ineffective assistance of counsel, set forth in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), is “applicable to ineffective-assistance claims arising out of the. plea process.” Hill v. Lockhart, 474 U.S. 52, 57, 106 S.Ct. 366, 88 L.Ed.2d 203 (1985). We must first ask whether “counsel’s assistance was reasonable considering all the circumstances.” Strickland, 466 U.S. at 688, 104 S.Ct. 2052. Counsel must have “wide latitude ... in making tactical decisions,” and “judicial scrutiny of counsel’s performance must be highly deferential.” Id. at 689, 104 S.Ct. 2052. We “must indulge a strong presumption that counsel’s conduct falls within the wide range of reasonable professional assistance.” Id. In the context of that presumption, we “must then determine whether, in light of all the circumstances, the identified acts or omissions were outside the wide range of professionally competent assistance.” Id. at 690, 104 S.Ct. 2052.
Second, if counsel’s performance was deficient, we assess prejudice. Prejudice “focuses on whether counsel’s constitutionally ineffective performance affected the outcome of the plea process.” Hill, 474 U.S. at 59, 106 S.Ct. 366. “In other words, in order to satisfy the ‘prejudice’ requirement, the defendant must show that there is a reasonable probability that, but for counsel’s errors, he would not have pleaded guilty and would have insisted on going to trial.” Id. (footnote omitted).
Under AEDPÁ, review of the state court’s application of Strickland is “doubly deferential” to the performance of counsel, because a petitioner must show that the state court’s ruling was an objectively unreasonable application of Strickland. 28 U.S.C. § 2254(d)(1); see also Mirzayance, 556 U.S. at 123, 129 S.Ct. 1411; Bell v. Cone, 535 U.S. 685, 698-99, 122 S.Ct. 1843, 152 L.Ed.2d 914 (2002).
1. It was not objectively unreasonable for the state PCR court to conclude that counsel’s performance was not deficient.
The state court did not unreasonably apply Strickland. Because Hedlund has not shown that his counsel performed deficiently in making the tactical decision to attempt to move Hedlund’s plea proceedings before a different judge, relief is unavailable. Hedlund’s arguments that counsel failed to present the second plea in a timely manner and that there was a *577reasonable probability that the trial judge would have accepted that plea are not supported by the record.14
First, on the day both counsel were supposed to appear in chambers to discuss the second plea agreement, Hedlund’s counsel called the court to ask informally whether the judge would recuse himself. Counsel explained that “Mr. Hedlund would be willing to enter into a plea agreement but not in front of Judge Sheldon.” The judge’s assistant responded that the judge would not recuse himself and since counsel did not appear that day as required, the court would no longer entertain further plea agreements. Based on Judge Sheldon’s response, Hedlund’s counsel filed a motion for change of judge for cause in which he challenged “the bias exhibited by the court with regard to Mr. Hedlund.” In the motion, counsel explained:
Hedlund is willing to enter into [the second] plea agreement in any court other than this court. Defendant Hedlund feels that this court has become biased against him. He feels that he will not be offered a realistic opportunity to persuade this court at the time of sentencing that any sentence other than the maximum consecutive sentence is appropriate. This feeling is based, in part, on the court[’]s sua sponte decision to impanel dual juries, the denial of all substantive pretrial motions filed by the defense and the court’s demeanor leading up to trial.... The court[’]s failure to recuse itself would be tantamount to forcing the death penalty upon defendant Hedlund. As the court is aware, there is a significant amount of evidence against Mr. Hedlund in these cases. It is Mr. Hedlund’s purpose to avoid the death penalty in this case.
At the motion hearing before another judge, Judge Sheldon testified regarding his concerns with the first plea agreement and the fact that a second plea agreement was never formally offered. When Hed-lund’s counsel examined Judge Sheldon, Judge Sheldon also explained that (1) he was concerned about the plea being commensurate with culpability, (2) he took into account victim letters received from McClain’s family, and (3) continuing the plea process when a plausible plea was not on the table would only waste time and thwart the arrangements for a single trial with dual juries.
In his closing remarks, defense counsel argued why he thought Judge Sheldon was biased and why it would result in an unfair trial for Hedlund. With respect to the plea process, counsel highlighted the fact that Hedlund refused to plead before Judge Sheldon. Specifically, counsel stated that Hedlund
would be willing to enter into a plea but not in front of that Court [Judge Sheldon]. He would be willing to enter it in front of any other Court and this is again, a plea Judge Sheldon would most likely have been amenable to, but Mr. Hedlund felt he would not get a fair shake and still the Court said, no, we will not recuse ourselves so let justice be done.
Counsel concluded with an impassioned argument about the justice system and the importance of maintaining the community perception of fairness to victims and defendants alike. Counsel pleaded he was not asking for a handout, but “[w]hat he [was] asking on behalf of [Hedlund] is fairness, *578the ability to be heard before a Court without the appearance of impropriety.”
In rebuttal, counsel argued that, when you put all of the things Judge Sheldon did together, “it is enough for [Hedlund] and I to believe for the community to say, hold it, he is not getting a fair shake. There is the appearance of impropriety in reading those [victim] letters at that time and not giving him the benefit of a presentence report.” Counsel argued that the letters were not merely victim letters, but ex parte communications from state witnesses who also happened to be victims. Counsel reiterated that rejection of the plea to facilitate moving forward with the dual jury procedure was also improper.
This record demonstrates that counsel’s motion to have Hedlund’s case moved before a different judge was purely a tactical decision.15 Counsel apparently honestly believed that Hedlund could not get a “fair shake” in front of Judge Sheldon. Even though counsel believed Judge Sheldon was likely to accept the second plea, counsel persisted with the request. He persisted, because he thought Hedlund faced an undue risk of bias and would surely receive a death sentence from Judge Sheldon if the second plea agreement were not accepted and the case proceeded to trial. Counsel’s written motion and arguments made clear that it was Hedlund’s primary goal at this point to avoid the death penalty. We must give deference to counsel’s tactical decision to do whatever he could to put his client in front of a non-biased judge (who was not pre-inclined to sentence Hed-lund to death). It was not error for the state PCR court to conclude that counsel’s performance was not deficient. Indeed, counsel made strong arguments about the judge having ex parte communication with the state’s witnesses (who were also victims) and gave many reasons for wanting the case moved before another judge.
Hedlund’s argument that counsel missed the deadline for the second plea agreement is a red herring. At base, this argument again challenges counsel’s tactical decision. On the day defense counsel and the prosecutor were supposed to appear in chambers to discuss the second plea agreement, counsel instead put the wheels of recusal in motion. He called chambers requesting recusal. When the judge declined, he proceeded with a formal motion to have the recusal motion heard before another judge so that the plea process could continue in front of an unbiased jurist and, without the dual jury deadline hanging over his head. This too was a tactical decision; it was not an act of ineompetency. Because counsel’s performance did not fall outside of the wide range of professionally competent advice, the state courts did not unreasonably apply the first prong of Strickland.
2. No prejudice has been shown.
Even assuming the state PCR court’s application of Strickland was objectively unreasonable, Hedlund has not shown a reasonable probability that, but for counsel’s errors, Hedlund would not have gone to trial. In other words, the record does not demonstrate that, if counsel would have presented the second plea agreement to Judge Sheldon (instead of calling chambers to ask for recusal), there is a reasonable probability Judge Sheldon would have accepted the agreement and Hedlund would have avoided the death penalty. See Lafler v. Cooper, 566 U.S. 156, 132 S.Ct. 1376, 1385, 182 L.Ed.2d 398 (2012) (“In these circumstances a defendant must show that but for the ineffective advice of counsel there is a reasonable *579probability that the plea offer would have been presented to the court ..., that the court would have accepted its terms, and that the conviction or sentence, or both, under the offer’s terms would have been less severe than under the judgment and sentence that in fact were imposed.”).
Although Hedlund argues that the second plea with respect to the McClain homicide would have complied with the range of acceptable penalties to which the trial court would have agreed, it is unlikely that the court would have accepted the plea as to either the Mertens or the McClain crimes.
First, with respect to the McClain homicide, while Judge Sheldon had indicated that first degree burglary would be a starting point, “[a]t that point, [Judge Sheldon] had not made up [his] mind whether or not that would be an appropriate disposition because [he] still ... continued to have serious reservations about the disposition of this case given the charges against [Hedlund].” Judge Sheldon testified with respect to the first plea agreement, “Quite frankly, I was very surprised there had not been a plea to First Degree Murder with the State stipulating it would not seek the death penalty, and I was surprised there had been a plea to Second Degree Murder and I think from what I gathered in [defense counsel’s] conversations, that [counsel] shared my reservations about being able to establish a factual basis for Second Degree Murder to a Felony Murder charge because the law is quite clear, there are no lesser included offenses to Felony Murder.” Based on the court’s statements, this plea would not have provided sufficient accountability for the McClain homicide. There is nothing else in the record suggesting a reasonable probability that the court would have accepted the new offer of a plea to theft with a prior and burglary non-dangerous with respect to McClain.
Second, with respect to Mertens, during the hearing on the change-of-judge motion, Judge Sheldon testified that, after reviewing the first plea agreement, he
continue[d] to have reservations about [the second degree murder plea for the Mertens homicide] and as I indicated to [defense counsel], at the conclusion of that hearing, that I was — [defense counsel] had indicated to me apparently [he] and [the prosecutor] were going to continue plea negotiations or try and work something out.
Judge Sheldon further testified that he “continued to have reservations as you all did in stating to me you weren’t sure whether or not a plea to Second Degree Murder, you would be able to establish a factual basis, so there were reservations ... between all parties at that point.” With respect to the first plea agreement, even after the parties recited a factual basis for second degree murder, the court’s concerns “were not dispelled” as to whether the plea could be accepted for the Mertens homicide. Again, there is nothing in the record to suggest that the court’s concerns would have been dispelled such that it would have accepted the second plea agreement’s identical offer of second degree murder for the Mertens crime.
Third, Judge Sheldon expressed concern about “disparate treatment given to ... co-defendants” and whether this would create due process concerns under existing Supreme Court precedent. Judge Sheldon also explained that, if it turned out Hed-lund was just as culpable or more culpable than McKinney, he would have been allowed less severe punishment under the plea agreement while McKinney faced the death penalty. Counsel was given the opportunity to explain during the informal plea discussion how Hedlund was less cul*580pable than McKinney, but the judge “simply did not hear it.”
In sum, Judge Sheldon expressed (1) ongoing reservations about even accepting a second degree murder plea for the Mer-tens homicide, (2) concern that the plea reflect the appropriate amount of culpability for the McClain homicide (given the strong evidence against Hedlund), and (3) a desire to avoid disparate sentences. Moreover, the record indicates that Hed-lund was not willing to enter a plea agreement in front of Judge Sheldon. When defense counsel called Judge Sheldon’s chambers asking the judge to recuse himself, the explanation defense counsel provided was that “Hedlund would be willing to enter into a plea agreement but not in front of Judge Sheldon.” He provided the same explanation in his motion to recuse. On this record, it cannot be said that, if Hedlund’s counsel had presented the second plea to Judge Sheldon, there is a reasonable probability it would have been accepted and the death penalty avoided. Thus, Hedlund has failed to show prejudice.
V. Ineffective Assistance of Counsel During Penalty Phase16
A. Background and procedural history
At trial, Hedlund presented expert testimony from Dr. Ronald Holler, who had conducted a “Neuropsychological and Psychological Evaluation” of Hedlund before trial. Dr. Holler noted that Hedlund reported drinking up to twelve beers on the night of the burglary-murder. He found that Hedlund’s intoxication was a function of his “alcohol dependence.” He then discussed in some detail Hedlund’s “extremely dysfunctional” early childhood experiences. Dr. Holler found that Hedlund had a “misguided loyalty” toward McKinney and had a limited understanding of his “personality inadequacies.” Regarding Hedlund’s “Intellectual/N europsychological Functioning,” he found a “low average” IQ. He also found Hedlund may have scored low on certain tests due to an “underlying depressive status” and that Hed-lund displayed “a slight indication of a learning disability.”
Dr. Holler “evaluate[d] various aspects of [Hedlund’s] intellectual, cognitive, neu-ropsychological, [and] emotional functioning as related to his background with his family and other aspects of his environment.” One of the tests Dr. Holler administered was the “Concise Neuropsychological Scale.” He focused on “the abuse [Hedlund] suffered and the resulting psy-choneurological effects” of that abuse. He opined that Hedlund suffered from “Post-traumatic Stress Disorder [PTSD], as well as some intertwined disorders of much consequence, including the alcohol dependence and a depressive disorder.” He explained how the psychological and physical abuse Hedlund suffered can lead to these disorders.
Specifically, Dr. Holler explained the “neuropsychological impairment” that can result and stated that Hedlund showed “some indications of a very significant but yet in a sense mild neuropsychological deficit.” Counsel then specifically inquired about brain damage.
Q: Did you find any indication of right hemisphere brain dysfunction or disorder?
A: There were indications of this. His verbal IQ was 91, performance IQ was 78. Essentially we talk about the verbal IQ as being primarily associated with *581left hemisphere functioning and this does refer then to receptive and expressive speech, reading capability and verbal memory.... [The test results provide] further evidence that the right hemisphere is not functioning as well as the left hemisphere. This may well be related to some of the physical abuse that he experienced, including being hit on the back of the head.
Dr. Holler went on to explain that damage to the right hemisphere could affect someone’s judgment. On redirect, he clarified that, while Hedlund was not “severely retarded” or “totally psychotic,” Hedlund did have “neurological impairments which impaired his judgment.”
Dr. Charles Shaw, a medical addiction specialist, also testified regarding Hed-lund’s alcoholism. He testified that alcoholism can lead to organic brain damage. He also believed that Hedlund’s actions with respect to the crimes were influenced by his alcoholism.
At sentencing, the trial court did not find credible evidence to support Dr. Shaw’s conclusion that Hedlund was affected by alcohol at the time of the crimes. Instead, the court found that Hedlund had a motive to lie about the extent of his alcohol consumption and his statements conflicted with those of his sisters and a presentence report from an earlier conviction.
The court also discounted Dr. Holler’s testimony, because (1) he did not raise PTSD in his initial report, instead announcing it for the first time while testifying; (2) some of the foundational information upon which Dr. Holler based his opinions was self-reported by Hedlund; and (3) some of the conclusions were based on an erroneous presentence report.
During PCR proceedings, Hedlund proffered a report from Dr. Marc S. Walter, a neuropsychologist. Dr. Walter conducted a battery of tests on Hedlund and found certain results consistent with a diagnosis of alcohol abuse. He also found “Cognitive Disorder, Not Otherwise Specified,” a disorder “that used to be termed Organic Mental Disorder and indicates the presence of brain damage,” and stated that Hedlund may have “residual problems” with PTSD. In light of these results, Dr. Walter concluded that Hedlund had brain damage at the time of the offenses in 1991.
Dr. Walter admitted that the test used by Dr. Holler was a “screening test for brain damage.” He expressed a preference, however, for the battery of tests he administered because they are a “comprehensive neuropsychological test battery.” Dr. Walter stated that screening tests such as those used by Dr. Holler “are relatively insensitive and often miss the presence of brain damage.” Dr. Walter concluded by stating that he believed that Hedlund’s brain damage, as augmented by his alcohol use, prevented Hedlund from “understanding] the consequences of his involvement in the burglaries and the murders.”
The PCR court reviewed Dr. Walter’s report but concluded that counsel’s efforts during sentencing did not fall below the standard expected of reasonable death-penalty trial lawyers. The court noted that Dr. Walter’s report would not support an insanity defense, and nothing in the record ■suggested Hedlund was unaware of his involvement in the crimes. The court continued that “[t]he fact that an attorney, after the fact, obtains an opinion from an expert which might have supported an alternative theory at trial does not demonstrate, without more, that the strategy chosen by defense counsel at the time of trial was ineffective.”
The court rejected the argument that counsel did not present sufficient evidence *582of the neuropsychological effects of Hed-lund’s child abuse and alcohol abuse. The court stated that it was adequately informed of these conditions by Drs. Holler and Shaw. The court found that Dr. Walter’s report was not substantially or significantly different from the earlier expert reports. The court challenged Dr. Walter’s conclusion that Holler did not diagnose brain damage, which in fact he did.
The district court reviewed all of the expert testimony and reports proffered during the penalty phase and in PCR proceedings. Based on that review, the court concluded that it was not objectively unreasonable for the PCR court to find that (1) the penalty phase experts’ opinions and PCR expert’s opinion were substantially the same, and (2) Dr. Holler entertained a diagnosis of brain impairment. The district court also found that the PCR court did not unreasonably apply Strickland. It rested this holding only on the performance prong, finding analysis of the prejudice prong unnecessary.
B. The state court did not unreasonably apply Strickland.
On federal habeas review of ineffective assistance of counsel claims, courts apply the clearly established federal law set forth in Strickland. See e.g., Cullen v. Pinholster, 563 U.S. 170, 189, 131 S.Ct. 1388, 179 L.Ed.2d 557 (2011). Under Strickland, we must first ask whether “counsel’s assistance was reasonable considering all the circumstances.” Strickland, 466 U.S. at 688, 104 S.Ct. 2052. Counsel is granted “wide latitude ... in making tactical decisions,” and “judicial scrutiny of counsel’s performance must be highly deferential.” Id. at 689, 104 S.Ct. 2052. We must also “indulge a strong presumption that counsel’s conduct falls within the wide range of reasonable professional assistance.” Id. In the context of that presumption, we “determine whether, in light of all the circumstances, the identified acts or omissions were outside the wide range of professionally competent assistance.” Id. at 690, 104 S.Ct. 2052.
Even a “professionally unreasonable” error by counsel will not warrant setting aside a judgment, unless it was “prejudicial to the defense.” Id. at 691-92, 104 S.Ct. 2052. To establish prejudice, a “defendant must show that there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different. A reasonable probability is a probability sufficient to undermine confidence in the outcome.” Id. at 694, 104 S.Ct. 2052.
The PCR court’s factual findings were not objectively unreasonable. The findings, that the reports of Drs. Holler and Shaw were substantially the same as Dr. Walter’s proffered report and that Dr. Holler diagnosed brain damage, are supported by the record. Dr. Holler found that Hedlund suffered from alcohol dependence, PTSD, and a depressive disorder. Dr. Holler also explained how neurological impairment can result from those factors, and that Hed-lund had indications of “a very significant but yet in a sense mild neuropsychological deficit.” Dr. Walter admitted that the test used by Dr. Holler screens for brain damage and Dr. Holler found that Hedlund had a right hemisphere dysfunction or disorder and that this could impair his judgment. Dr. Shaw testified about Hedlund’s alcoholism and its effects on Hedlund. Similarly, Dr. Walter opined about brain damage and its impact at the time of the offense.
1. It was not objectively unreasonable for the state PCR court to conclude that counsel’s performance was not deficient.
The PCR court’s application of Strickland was also not unreasonable. Hedlund’s *583counsel’s performance was reasonable considering the circumstances. Counsel hired a psychologist to testify about Hedlund’s various mental and personality defects, including neuropsychological impairments to his brain. Counsel also hired a psychiatrist to testify about Hedlund’s severe alcoholism. Counsel’s tactical decisions of precisely which experts to hire must be afforded deference. Hedlund’s proffer of additional experts on collateral review who say substantially the same thing does not call into question the reasonableness of counsel’s decisions. Counsel’s strategy to present testimony about Hedlund’s troubled childhood and ongoing psychological, neuropsy-chological, and medical conditions cannot be said to fall outside the wide range of professionally competent assistance.
Hedlund argues that the PCR court contradicted itself with respect to the expert testimony presented during sentencing. Specifically, on PCR review, the court found testimony by Drs. Holler and Shaw sufficient to paint a picture of Hedlund’s condition. However, Hedlund argues that when sitting as the sentencing court, the court discredited the same experts’ testimony.
That the sentencing court discredited certain aspects of Drs. Holler and Shaw’s testimony does not discredit the PCR court’s conclusion that their opinions were substantially the same as that proffered by Dr. Walter. During sentencing, the court discredited Dr. Shaw’s conclusion that Hedlund was affected by alcohol at the time of the crimes. The court found this self-reported information suspect, because of Hedlund’s motive to lie. The court also questioned why Dr. Holler raised PTSD for the first time while testifying — when he had not cited it in his report — and noted that some of the conclusions were based on erroneous information contained in a presentence report. These observations do not call into question Dr. Shaw’s conclusion that Hedlund suffered from alcoholism or Dr. Holler’s conclusion that Hedlund suffered from a brain impairment. They simply speak to the weight afforded the experts’ opinions in determining mitigation — weight based on reliability and credibility. To the extent Dr. Walter’s testimony also relied on the sentencing transcript, reports from family members, and information self-reported by Hedlund, it would be unreliable for the same reasons.
Hedlund also argues that counsel did not have “a complete picture” of his brain damage and, if counsel would have hired a neuropsychology expert, the expert could have “definitively concluded” that Hedlund had brain damage. However, as explained above, the PCR court did not make objectively unreasonable factual determinations that evidence of brain damage presented at sentencing was similar to that proffered to the PCR court. Hedlund has also failed to rebut the presumption that counsel’s preparation of the expert witnesses for sentencing fell below the wide range of professionally acceptable conduct.
2. Prejudice
Because Hedlund has not shown that counsel’s performance was deficient, we need not reach the question of prejudice.
VI. Consideration of Mitigating Evidence Under Lockett/Eddings 17
A. Background and procedural history
During the penalty phase of trial, the trial court found evidence of Hedlund’s *584tortured childhood to be compelling and credible. However, the court found that the mitigating factors (Hedlund’s childhood abuse and long-term alcohol use) did not outweigh the aggravating factors. The court reached this conclusion because, at the time of the crime, these factors did not affect Hedlund’s behavior or prevent him from knowing right from wrong. The trial court thus sentenced Hedlund to death.
When the Arizona Supreme Court conducted an independent review of the mitigating factors, it struck one of Hedlund’s aggravating factors and reweighed the remaining aggravating factor against the mitigating evidence. The court then found that the aggravating factor was not overcome.
' The federal district court also found that Hedlund’s trial court fulfilled its duty to consider all of the mitigating evidence and that it did not impose a relevancy test “or any other barrier” to consideration of this evidence. The district court concluded that no constitutional error arose when the trial court assigned less weight to the family background and alcohol mitigating evidence because it did not influence Hed-lund’s criminal conduct.
B. The Arizona Supreme Court applied an unconstitutional causal nexus test to Hedlund’s mitigating evidence.
We now consider whether the Arizona Supreme Court applied an unconstitutional causal nexus test in affirming Hedlund’s death sentence on its independent review of Hedlund’s death sentence. We first look at our precedent regarding the role mitigation evidence plays in sentencing decisions. We then apply our recent decision in McKinney v. Ryan, 813 F.3d 798 (9th Cir.2015) (en banc).
1. A sentencing court may not refuse to consider any relevant mitigating evidence.
In Lockett v. Ohio, 438 U.S. 586, 98 S.Ct. 2954, 57 L.Ed.2d 973 (1978), the Supreme Court held:
[T]he Eighth and Fourteenth Amendments require that the sentencer ... not be precluded from considering, as a mitigating factor, any aspect of a defendant’s character or record and any of the circumstances of the offense that the defendant proffers as a basis for a sentence less than death....
Given that the imposition of death by, public authority is so profoundly different from all other penalties, ... [the sentencer must be free to give] independent mitigating weight to aspects of the defendant’s character and record and to circumstances of the offense proffered in mitigation....
Id. at 604-05, 98 S.Ct. 2954 (finding Ohio death penalty statute invalid where it permitted consideration of only three mitigating circumstances).
Later, in Eddings v. Oklahoma, the Supreme Court applied Lockett in a capital case where the trial judge stated that he could not consider mitigating evidence of the defendant’s family history.18 455 U.S. 104, 112-13, 102 S.Ct. 869, 71 L.Ed.2d 1 (1982). The appeals court affirmed the trial court, finding that the mitigation evidence was “not relevant because it did not tend to provide a legal excuse” from criminal *585responsibility. Id. at 113, 102 S.Ct. 869. The Supreme Court reversed, explaining:
Just as the State may not by statute preclude the sentencer from considering any mitigating factor, neither may the sentencer refuse to consider, as a matter of law, any relevant mitigating evidence .... The sentencer ... may determine the weight to be given relevant mitigating evidence. But [it] may not give it no weight by excluding such evidence from [its] consideration.
Id. at 113-15, 102 S.Ct. 869.19
For a period of a little over 15 years, in violation of Eddings, the Arizona Supreme Court articulated and applied a “causal nexus” test in capital cases. The test forbade giving weight to nonstatutory mitigating evidence, such as family background, unless such evidence was causally connected to the crime.20 See, e.g., State v. Wallace, 160 Ariz. 424, 773 P.2d 983, 986 (1989) (en banc) (“A difficult family background is a relevant mitigating circumstance if a defendant can show that something in that background had an effect or impact on his behavior that was beyond the defendant’s control.”); accord State v. Ross, 180 Ariz. 598, 886 P.2d 1354, 1363 (1994) (en banc) (“A difficult family background is not a relevant mitigating circumstance unless ‘a defendant can show that something in that background had an effect or impact on his behavior that was beyond the defendant’s control.’ ” (quoting Wallace, 773 P.2d at 986)).
In Tennard v. Dretke, the Supreme Court rejected a “nexus test” that would find mitigating evidence relevant only where it bears a causal nexus to the crime. 542 U.S. 274, 287, 124 S.Ct. 2562, 159 L.Ed.2d 384 (2004) (“[W]e cannot countenance the suggestion that low IQ evidence is not relevant mitigating evidence ... unless the defendant also establishes a nexus to the crime.”).21 In Smith v. Texas, the Court again considered the úse of a nexus test to determine whether any mitigating evidence is relevant. 543 U.S. 37, 45, 125 S.Ct. 400, 160 L.Ed.2d 303 (2004) (per curiam). The Court “unequivocally rejected” any test requiring a causal nexus between mitigating evidence and the crime. Id. We have held that Tennard and Smith are retroactively applicable to decisions such as the Arizona Supreme Court’s 1996 decision in this case. See Schad v. Ryan, 671 F.3d 708, 723 (9th Cir. 2009) (per curiam), overruled on other grounds by McKinney, 813 F.3d at *17.
In the past, to determine whether the Arizona Supreme Court used its causal nexus test, we applied a “clear indication” rule: We could find Eddings error only if there was a clear indication in the record that the court had refused, as a matter of law, to treat nonstatutory mitigation evi*586dence as relevant unless it had some effect on the petitioner’s criminal behavior. See Schad, 671 F.3d at 724. However, in McKinney, we determined that the “clear indication” rule was an “inappropriate and unnecessary gloss on the deference already required under § 2254(d).” 813 F.3d at *17.22
2. Application of the causal nexus test in this case.
The question (whether the Arizona Supreme Court applied the unconstitutional causal nexus test in sentencing Hedlund) has already been answered in the affirmative by our en banc court in McKinney, 813 F.3d at *17-20. As companion cases, the Arizona Supreme Court reviewed the death sentences of both Hed-lund and McKinney in the same opinion. See McKinney, 917 P.2d at 1214. In doing so, the Arizona Supreme Court intertwined its analysis for both Hedlund and McKinney, requiring the same outcome regarding this issue. Because we are bound by our court’s decision in McKinney, we follow its conclusion that the Arizona Supreme Court applied the unconstitutional causal nexus test in affirming Hedlund’s sentence.
The Arizona Supreme Court used much of the same reasoning in affirming the sentences for Hedlund and McKinney. First, the court cited to its prior opinion in Ross to support its conclusion that Hed-lund’s and McKinney’s difficult family background and childhood abuse did not necessarily have substantial mitigating weight. See McKinney, 917 P.2d at 1227 (Hedlund); id. at 1234 (McKinney). In McKinney, we noted regarding Hedlund’s sentence:
The [Arizona Supreme Court] first affirmed Hedlund’s death sentence, writing, “A difficult family background, including childhood abuse, does not necessarily have substantial mitigating weight absent a showing that it significantly affected or impacted a defendant’s ability to perceive, to comprehend, or to control his actions. See State v. Ross, [180 Ariz. 598] 886 P.2d 1354, 1363 (1994).” McKinney, 917 P.2d at 122[7]. As we pointed out above, the pin citation to Ross is a citation to the precise page on which the Arizona Supreme Court had two years earlier articulated its unconstitutional “causal nexus” test for non-statutory mitigation.
813 F.3d at *18. Later, when discussing McKinney’s sentence, we referred back to the Arizona Supreme Court’s analysis of Hedlund’s sentence, in which it said: “As we noted in discussing Hedlund’s claim on this same issue, a difficult family background, including childhood abuse, does not necessarily have substantial mitigating weight absent a showing that it significantly affected or impacted the defendant’s ability to perceive, comprehend, or control his actions. See State v. Ross, [180 Ariz. 598] 886 P.2d 1354, 1363 (1994)[J” Id. at *19 (alterations in original) (quoting McKinney, 917 P.2d at 1234).
Second, the Arizona Supreme Court adopted the sentencing court’s analysis of the mitigation evidence for both Hedlund and McKinney. For Hedlund, the sentencing court determined “that none of [Hed-lund’s] mitigating factors considered separately or cumulatively indicates to the Court that these factors affected the defendant’s ability to control his physical behavior at the time of the offense or to *587appreciate the wrongfulness of his conduct.” For McKinney, the sentencing court similarly found that the mitigation evidence did not “in any way affect[ ] [McKinney’s] conduct in this case.” As we explained in McKinney, “[The sentencing court’s] language ... .echoes the language of Arizona’s statutory mitigator under Ariz. Rev. Stat. § 13-703(G)(1). It also echoes the language used by the Arizona Supreme Court to articulate the unconstitutional causal nexus test applied to non-statutory mitigation.” McKinney, 813 F.3d at *18.
Thus, in McKinney, we concluded that the Arizona Supreme Court’s decision was contrary to Eddings, based in part on (1) “the Arizona Supreme Court’s recital of the causal nexus test for nonstatutory mitigation and its pin citation to the precise page in Ross where it had previously articulated that test,” and (2) “the factual conclusion by the sentencing judge, which the Arizona Supreme Court accepted, that McKinney’s [mitigation evidence] did not ‘in any way affect[] his conduct in this case.’ ” Id. at *20 (second alteration in original). This same reasoning applies to the Arizona Supreme Court’s decision for Hedlund. Accordingly, we adopt our en banc court’s conclusion in McKinney that the Arizona Supreme Court’s decision of Hedlund’s claims was contrary to Ed-dings.23
Having determinad that the Arizona Supreme Court committed Ed-dings error, we next must decide whether such error was harmless. See id. The harmless error standard on habeas review provides that “relief must be granted” if the error “had substantial and injurious effect” on the sentencing decision. See Brecht, 507 U.S. at 623, 113 S.Ct. 1710 (quoting Kotteakos, 328 U.S. at 776, 66 S.Ct. 1239); McKinney, 813 F.3d at *21. Again, we adopt our conclusion in McKinney. The Eddings error (committed by the Arizona Supreme Court in this case) had a “substantial and injurious effect” on Hed-lund’s sentence within the meaning of Brecht, and was, therefore, not harmless. See McKinney, 813 F.3d at *21-22.
CONCLUSION
The district court properly denied relief on Hedlund’s claims regarding (1) use of the visible leg brace, (2) use of dual juries, (3) juror bias, (4) ineffective assistance of counsel during the plea process, and (5) ineffective assistance of counsel during the penalty phase. However, the district court should have granted the petition with respect to Hedlund’s sentence, based on Hedlund’s claim regarding (6) the Arizona Supreme Court’s consideration of mitigating evidence under Lockett, Eddings, and their progeny. Accordingly, we reverse the district court’s judgment denying the writ of habeas corpus. We remand with instructions to grant the writ with respect to *588Hedlund’s sentence unless the state, within a reasonable period, either corrects the constitutional error in his death sentence or vacates the sentence and imposes a lesser sentence consistent with law.
Each party shall bear its own costs on appeal.
REVERSED in part, AFFIRMED in part, and REMANDED.
. Judge Wardlaw dissents from Parts I, IV, and V of this disposition, stating that she has “previously explained [her] disagreement with the majority's disposition of Hedíünd’s claims of unconstitutional shackling during trial and ineffective assistance of counsel during the plea process and penalty phase.” Op. at 592 (Wardlaw, J., concurring in part and dissenting in part) (citing Hedlund v. Ryan, 750 F.3d 793, 831-43 (9th Cir. 2014) (Ward-*563law, J., concurring in part and dissenting in part)). In our prior opinion, we responded to her disagreement. Hedlund, 750 F.3d at 811 n.15, 811-12, 813 n.16, 817, 820, 823 n.25. Similar to Judge Wardlaw’s statement, we see no need to repeat our disagreement with her prior dissent here.
. Because Hedlund has not shown that resolution of his remaining claims is "debatable amongst jurists of reason,” Miller-El v. Cockrell, 537 U.S. 322, 336, 123 S.Ct. 1029, 154 L.Ed.2d 931 (2003), we decline to reach the other uncertified issues on appeal. See 28 U.S.C. § 2253(c); Hiivala v. Wood, 195 F.3d 1098, 1102-04 (9th Cir. 1999) (per curiam).
. When Lane was recalled al a later time, he testified that Hedlund’s "jail card” (which tells officers about the risks posed by inmates), contained a narrative about an escape plan. Specifically, the narrative read, "Warning, take keys and clothing per class A1920. McKinney planning escape by jumping guard per information, 300120, per request CPD 2525.” While no specific mention of Hedlund was given in this narrative, the escape warning was presumably applied to him as well because the narrative appeared on Hedlund's jail card.
. Although the leg restraint was intended to be invisible, the record demonstrates that it was in fact visible to the jury. Indeed, Respondent conceded visibility at oral argument. Insofar as the restraints were visible, however, the trial court found Hedlund largely to blame. In particular, it found that "had [he] chosen to do so, [Hedlund] could have facilitated the concealment of the leg brace by keeping [his] pants pulled down, and [his] legs back from the front of the desk.”
. As fully discussed below, this recitation of the facts is in error. The record shows that it was McKinney, not Hedlund, who attempted an escape in 1991.
. Even if we assume that the Arizona Supreme Court’s erroneous factual statement (misattributing the 1991 escape attempt to Hedlund) is enough to call into question the entirety of the factual findings regarding shackling, conducting de novo review would not change the outcome.
. Where the decision to physically restrain a defendant violates due process, on habeas review, a petitioner must show that the error had "substantial and injurious effect or influence in determining the jury’s verdict.” Brecht v. Abrahamson, 507 U.S. 619, 623, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993) (quoting Kotteakos v. United States, 328 U.S. 750, 776, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946)). "To determine whether the imposition of physical restraints constitutes prejudicial error, we have considered the appearance and visibility of the restraining device, the nature of the crime with which the defendant was charged and the strength of the state’s evidence against the defendant.” Larson v. Palmateer, 515 F.3d 1057, 1064 (9th Cir. 2008) (emphasis added). However, we have also recognized that this multi-factor test is not clearly established federal law. Walker v. Martel, 709 F.3d 925, 938 (9th Cir. 2013). In any event, because we find that the use of a leg restraint did not violate due process, we do not reach the issue of prejudice.
.The court arranged for this evidence to be heard separately to avoid a possible Bruton v. United States, 391 U.S. 123, 88 S.Ct. 1620, 20 L.Ed.2d 476 (1968), violation. In Brnton, during a joint trial, the trial court instructed the jury that a codefendant’s confession inculpating both the codefendant and the defendant could be used only against the codefendant, and should be disregarded with respect to the defendant. Id. at 124-25, 88 S.Ct. 1620. Where the jury was allowed to consider the codefendant’s confession, the Supreme Court found that the confession "added substantial, perhaps even critical, weight to the Government’s case in a form not subject to cross-examination, since [the codefendant] did not take the stand. [The defendant] thus was denied his constitutional right of confrontation." Id. at 128, 88 S.Ct. 1620. The Court recognized that "[t]he unreliability of [inculpatory statements by a codefendant] is intolerably compounded when the alleged accomplice, as here, does not testily and cannot be tested by cross-examination.” Id. at 136, 88 S.Ct. 1620. The Court concluded that "in the context of a joint trial we cannot accept limiting instructions as an adequate substitute for petitioner’s constitutional right of cross-examination.” Id. at 137, 88 S.Ct. 1620.
. Those procedures included separate voir dire of the jury panels, a courtroom layout that allowed both Defendants full view of the jurors and witnesses, separate preliminaiy instructions, separate opening statements, separate reading of the charges, special procedures for handling codefendant inculpatory statements, separate closing statements, and special procedures for the return of the verdicts.
. At the same time, the Arizona Supreme Court also reversed its earlier decision in State v. Lambright, 138 Ariz. 63, 673 P.2d 1 (1983) (en banc), which had found that the use of dual juries violated state law.
. Citing United States v. Gonzalez, 214 F.3d 1109, 1114 (9th Cir. 2000), Hedlund argues that the Juror’s statement was "somewhat equivocal.” In Gonzalez, we noted the difference between a juror who is somewhat indirect in their responses (e.g. Q: “Would your husband's experience keep you from serving impartially?” A: "I don’t believe so, no.”; Q: "Could you set aside your feelings and act impartially?” A: "I believe so, yes.”), and a juror who answers equivocally three times in a row to whether she could be fair ("I will try to”; "Right. I’ll try”; and "I'll try”). Id. at 1111, 1114. We recognized that it would be acceptable to retain the first juror, because after stating her belief, the juror followed up with "an unqualified affirmative or negative” regarding impartiality. Id. at 1114. The same can be said for the Juror. In her letter, she initially stated "I don’t believe it would affect my ability to be fair and impartial,” then when questioned by the trial court, she added "an unqualified affirmative” when she was asked to confirm her belief that she could be impartial (Q: "You state here at the end that, T don’t believe it would affect my ability to be fair and impartial.’ Is that your belief?” A: "Yes, it is.”). While the trial court asked the question somewhat inartfully, the Juror’s response does not display equivocation. Moreover, the trial court credited her response after asking further questions, observing her demeanor, and judging her credibility. This finding is entitled to a presumption of correctness. Rushen v. Spain, 464 U.S. 114, 120, 104 S.Ct. 453, 78 L.Ed.2d 267 (1983).
. Although not controlling, Justice O'Con-nor’s concurrence in Smith expressed concern about cases involving juror misconduct. Therein, she listed certain "extreme situations” in which she believed a bias hearing may be inadequate and implied bias could be found. Examples may include: "a revelation that the juror is an actual employee of the prosecuting agency, that the juror is a close relative of one of the participants in the trial or the criminal transaction, or that the juror was a witness or somehow involved in the criminal transaction.” Smith, 455 U.S. at 222, 102 S.Ct. 940 (O’Connor, J., concurring) (emphasis added). Because she read the majority opinion as not foreclosing the use of implied bias in certain situations, Justice O'Connor concurred. Id. at 224, 102 S.Ct. 940.
Even if this concurrence could be construed as clearly established federal law, the notion that implied bias could be found when a juror is a close relative does not lead to the conclusion that implied bias should be found when the juror is a former distant relative by virtue of two marriages, one now dissolved and the former relative now deceased. Moreover, Hedlund does not allege juror misconduct in this case. The Juror was forthcoming as soon as she found out about the former relation and there is no indication she tried to conceal bias to influence the outcome of the trial.
. The district court declined to grant a COA on this issue. However, because we conclude that the district court's resolution of the issue is "debatable amongst jurists of reason,” Miller-El, 537 U.S. at 336, 123 S.Ct. 1029, we address it.
. As an initial matter, it is not clear that the second offer was still valid at the time in question. According to the prosecutor, Hed-lund rejected the second plea offer two days before defense counsel called chambers and asked the judge to recuse himself.
. With respect to preserving the plea in the record, counsel set forth the terms of the plea in his written motion and explained the terms of the plea at the motion hearing.
. The district court declined to grant a COA on this issue. However, because we conclude that the district court’s resolution of the issue is "debatable amongst jurists of reason,” Miller-El, 537 U.S. at 336, 123 S.Ct. 1029, we address it.
. The district court declined to grant a COA on this issue. However, because we conclude that the district court's resolution of the issue is “debatable amongst jurists of reason,” Miller-El, 537 U.S. at 336, 123 S.Ct. 1029, we address it.
. In Eddings, the sentencing judge made clear, on the record, that he could not consider certain evidence as a matter of law. He stated: "[T]he Court cannot be persuaded entirely by the ... fact that the youth was sixteen years old when this heinous crime was committed. Nor can the Court in following the law, in my opinion, consider the fact of this young man’s violent background." 455 U.S. at 109, 102 S.Ct. 869 (alterations in original).
. The Court later explained that "Eddings makes clear that it is not enough simply to allow the defendant to present mitigating evidence to the sentencer. The sentencer must also be able to consider and give effect to that evidence in imposing sentence.” Penry v. Lynaugh, 492 U.S. 302, 319, 109 S.Ct. 2934, 106 L.Ed.2d 256 (1989), abrogated on other grounds by Atkins v. Virginia, 536 U.S. 304, 122 S.Ct. 2242, 153 L.Ed.2d 335 (2002).
. Arizona law provides five statutory mitigating factors, as well as a catchall nonstatutory mitigating factor encompassing “any factors proffered by the defendant or the state that are relevant in determining whether to impose a sentence less than death, including any aspect of the defendant's character, propensities or record and any of the circumstances of the offense.” Ariz. Rev. Stat. § 13-751(G). Ed-dings and Lockett apply only to nonstatutory mitigating evidence. See McKinney, 813 F.3d at *9.
.Following the Supreme Court’s decision in Tennard, the Arizona Supreme Court abandoned its causal nexus test. See State v. Newell, 212 Ariz. 389, 132 P.3d 833, 849 (2006) (en banc); State v. Anderson, 210 Ariz. 327, 111 P.3d 369, 391-92 (2005) (en banc).
. We express no opinion as to how to apply McKinney in future Arizona capital cases from the suspect time period.
. We note that a court is free to assign less weight to mitigating factors that did not influence a defendant's conduct at the time of the crime. See Schad, 671 F.3d at 723 ("The United States Supreme Court has said that the use of the nexus test in this manner is not unconstitutional because state courts are free to assess the weight to be given to particular mitigating evidence.”). However, a court may not refuse to consider mitigating evidence because it lacked a causal nexus to the crime. In sum, a court may consider causal nexus in assessing the weight of mitigating evidence, but not in assessing its relevance. The Arizona Supreme Court has correctly recognized this in post-Tennard cases. See Newell, 132 P.3d at 849 ("We do not require that'a nexus between the mitigating factors and the crime be established before we consider the mitigation evidence. But the failure to establish such a causal connection may be considered in assessing the quality and strength of the mitigation evidence.” (citation omitted)).