Daniel WILSON, Petitioner-Appellant, v. Betty MITCHELL, Warden, Respondent-Appellee, 498 F.3d 491

COLE, J., delivered the opinion of the court in which, CLAY, J., joined. ROGERS, J. (p. 515), delivered a separate opinion concurring in the result. OPINION R. GUY COLE, JR., Circuit Judge. Petitioner-Appellant Daniel Wilson seeks habeas relief from his conviction and death sentence for the murder of Carol Lutz. After a night of drinking, Wilson put Lutz into the trunk of her car and eventually set the car on fire, killing her.

The jury convicted Wilson and found three capital specifications making him eligible for the death penalty: (1) murder committed to escape detection for kidnapping; (2) murder during kidnapping; and (3) murder during aggravated arson. The prosecution proceeded to the penalty phase relying on only the first (evading-kidnapping) specification as an aggravator, and the jury sentenced Wilson to death.

Wilson now raises five claims for habeas relief; the strongest is his claim that the trial court’s instruction regarding his intoxication defense improperly shifted to Wilson the burden to disprove the knowledge element of the evading-kidnapping aggravator. We conclude, for reasons other than those relied on by the district court, that any error in this regard was harmless. We further conclude that Wilson’s remaining claims are without merit. Accordingly, we AFFIRM the district court’s denial of habeas relief. I. BACKGROUND A.

Facts In Elyria, Ohio, on Saturday, May 4, 1991, around 1:30 p.m., Wilson killed Carol Lutz by locking her in the trunk of her car, puncturing the gas tank, and setting the car on fire. Wilson then walked away, allowing Lutz to burn to death. On the previous afternoon, Wilson was drinking at the Empire Tavern, a bar he frequented. Between 5:00 and 6:00 p.m., he went to the home of Angie Shelton, a girl he dated, and an argument ensued. As they argued, Wilson got angry, slammed her against the wall, threw her on the bed, and went to hit her.

Shelton told him that if he hit her, she “would be the last person that he hit.” Wilson then left, and later returned to the Empire Tavern. That evening, Carol Lutz drove her 1986 Oldsmobile Cutlass to the Empire Tavern to meet Douglas Pritt, an old boyfriend, and Wilson, apparently a new friend. Pritt, Lutz, and Wilson played pool and drank together. Pritt left the bar sometime between 12:30 a.m. and 1:00 a.m. Lutz and Wilson left close to 2:30 a.m. According to Wilson’s confession, Lutz offered him a ride home. She drove with him to the trailer where he lived.

Once there, she came in with him and they drank a couple of beers. Wilson vaguely recalled driving to Lorain, Ohio, to search for a party and stopping at his father’s house. Darlene DeBolt, a service-station cashier in Stow, Ohio, stated that Wilson stopped at the station around 5:55 a.m. on May 4. He was driving a black Oldsmobile Cutlass and appeared to be alone. DeBolt did not hear any suspicious noise coming from the Oldsmobile. Wilson told DeBolt, an old friend, that the car was his, that he had just driven from Canada, and that he “stopped a few states back for a few beers.” DeBolt smelled alcohol on him.

Wilson tried to get DeBolt to go out with him and was “persistent and pushy.” DeBolt refused to leave work and, after sixty or ninety minutes, Wilson left. When Wilson woke up later that morning, around 7:30 or 8:00 a.m., he was in a parking lot, sitting in the driver’s seat of Lutz’s Oldsmobile. Lutz, who was locked in the trunk, asked him to let her out, but he refused. Wilson could not recall how she got there. He drove to various places, including a park where he took a walk.

Wilson stated that he remembered thinking, “How am I going to get out of this?” Throughout this time, Lutz remained locked in the trunk. Still later that morning, Wilson drove to a school and parked the Oldsmobile. After a while he took off the gas cap, stuffed a rag in the open neck of the gas tank, and lit the rag, but the fire burned out. Lutz told him “she really had to go to the bathroom.” He “took the rag back out” of the gas tank and “let her [out to] go to the bathroom.” When he “told her to get back” in the trunk, “she stood there — she begged and pleaded with LWilson].

She begged — she’d turn around for 30 seconds and let [Wilson] run like hell.” Lutz told Wilson, “she’d go home and forget about it.” Wilson did not believe her and thought to himself, “How can you forget about being locked in a trunk?” Wilson stated that he did not leave her in the trunk because he “figured somebody would find her.... She’d get out and tell who I was.” When Wilson told her to get back in the trunk a second time, she complied. She sat in the trunk for fifteen to twenty minutes with the lid up.

They talked, and Wilson said “she asked me why don’t I just let her go?” He “even gave her a cigarette.” Then he closed the trunk lid, “poked a hole in the gas tank,” stuffed a towel or blanket into the gas tank, “let it soak with gas ... and ... lit it.” Then he “walked away from the car” and went to a nearby park. While out driving that day, Janette Patton and her mother noticed smoke and saw Lutz’s Oldsmobile enveloped in fire. After fire personnel extinguished the fire, they forced open the Oldsmobile’s trunk, revealing Lutz’s body. She died from third-degree .burns and carbon monoxide poisoning.

An arson investigator estimated that the flames could have heated the trunk to over 550 degrees, which could cause combustibles there to ignite and catch fire. There were no holes in the trunk, but there was a puncture in the gas tank. Investigators found a gas cap under the driver’s seat and a tire iron and cross bar in the back seat. Several samples of materials taken from inside the car tested positive for kerosene. Police detective Ray Riley traced the car to Carol Lutz and learned that she had last been seen with Wilson at the Empire Tavern. On May 9, police took Wilson into custody.

Riley interviewed Wilson after advising him of his Miranda rights. Wilson waived his rights and agreed to talk with the police. Riley tape recorded the interview. Wilson confessed to locking Lutz in the Oldsmobile’s trunk intermittently from 7:30 a.m. on May 4 until the time of her death. And he admitted that at approximately 1:30 p.m., he killed her by setting the Oldsmobile ablaze. The grand jury indicted Wilson on three aggravated-murder counts.

Count I charged aggravated murder by prior calculation and design; Count II charged felony murder, predicated on kidnapping; and Count III, as amended, charged felony murder, predicated on aggravated arson. Each murder count had three death specifications, which if found by the jury would make Wilson eligible for the death penalty. Specification one charged murder to escape “detection, apprehension, trial, or punishment” for kidnapping; specification two charged murder during kidnapping; and specification three charged murder during an aggravated arson.

Wilson was also indicted for kidnapping (Count IV) and aggravated arson (Count V). Wilson defended himself at trial by claiming intoxication and lack of prior calculation and design. The jury found Wilson guilty on all counts. At the penalty phase, the prosecutor elected to proceed to sentencing on only Count I (aggravated murder by prior calculation and design) and specification one (evading detection or punishment for another offense (kidnapping) in violation of Ohio Revised Code (O.R.C.) § 2929.04(A)(8)).

Accordingly, neither the court nor jury considered the other two murder counts or the felony-murder death-penalty specifications in assessing the penalty. At the outset of the defense’s penalty-phase case, a forensic toxicologist explained the effect that alcoholism has on a person’s body, mind, and behavior. Wilson’s mother, younger brother, grandfather, and aunt testified as to Wilson’s childhood. The witnesses explained that Wilson’s alcoholic father brutalized his wife and three sons throughout Wilson’s childhood.

Wilson’s father would lock his sons in their bedroom at night and refuse to let them out, even to go to the bathroom. The father teased and belittled his sons. In drunken rages, Wilson’s father would call his sons, “liars, cheats, and thieves,” accuse them of stealing things he could not find, and hit them on their bare backsides with a leather belt. Wilson’s mother, Linda Wilson, testified that her husband frequently slapped and terrorized her. When Wilson was twelve, he was arrested for vandalizing a friend’s house. A year or so later, his mother moved out of the family home.

She took Wilson’s younger brother Donald with her, and left Wilson and his other brother David with their father. Wilson’s father did not properly care for his sons, even failing to buy food. Wilson and David were forced to steal to survive. They regularly broke into neighbors’ homes to steal food or money. When he was fourteen years old, Wilson broke into an elderly neighbor’s home. When the neighbor surprised him, Wilson struck the elderly man, causing him to fall and break his hip. Wilson then ripped the phone cord out of the wall and left.

The neighbor was not found for two days and died as a result of his injuries and the lack of medical attention. A juvenile court adjudged Wilson delinquent by reason of involuntary manslaughter and remanded him to the custody of the Ohio Department of Youth Services. Wilson spent one year in a state facility for serious offenders, and then was sent to a halfway house. He fared well at both facilities. Although Wilson was initially reluctant to accept responsibility for his neighbor’s death, he eventually did. Days before turning seventeen years old, Wilson went to live with Shirley Spinney, a foster parent.

Wilson adjusted well to living with Spinney. He graduated from high school, with a B average, and worked part-time while in school. After high school, Wilson continued to live with Spinney even after his release from the custody of Youth Services. Wilson attended college for two semesters while continuing to work. Spinney described Wilson as incredibly compassionate, sensitive, and considerate. Ultimately, Spinney discovered that Wilson had a serious drinking problem. At times, he went out and got very drunk, would telephone her, and she would pick him up and take him home.

In 1988, Spinney’s other foster child, Mark, was killed in an accident. Wilson was devastated by Mark’s death and he began to drink more heavily. Wilson’s girlfriend noted that Mark’s death had a strong impact on Wilson and that he seemed like a different person when he was drinking. The next year, Wilson left Spinney’s home to live with friends. He next moved in with his mother and grandfather, sleeping in a camper behind their house. While there, he attempted to expunge his juvenile record and made plans to join the Navy.

In an unsworn statement presented at trial, Wilson asserted that his father “could do no wrong” in Wilson’s eyes. In spite of all the terrible things his father had done, he liked his father and spent a lot of time with him. Wilson described his juvenile arrest and his incarceration. He also described Spinney’s positive influence and the progress he made while living with her. Wilson stated that after Mark died, Wilson gave up on life.

He denied that he “intended to hurt” Lutz, and said, “I still do not know why I reacted the way I did.” He “would like to say to her family [he is] sorry.” Wilson said he did not want to die, and asked for another chance at life. Wilson’s expert witness, psychologist James Eisenberg, examined Wilson and concluded that Wilson is above average in intelligence and has difficulty becoming emotionally involved with others.

Further, Wilson’s lifestyle was marked by “strong dependency needs, maladjustment, and chaos.” He suffered from alcohol dependence and a “mixed personality disorder with borderline and antisocial features.” According to Eisenberg, Wilson was the product of a “classic dysfunctional family marked by physical, emotional and psychological abuse,” but he still identified with his father, not his battered mother. Wilson knew right from wrong, his ability to conform to the law was not impaired, and he could adjust and function in an institutional setting.

In rebuttal, Martha Lutz, Carol’s mother, testified about the devastating impact of Carol’s death on the family. They had been very close and had done many things together, including,shopping together frequently. Martha stated that she has a “broken heart that’s never going to heal” and misses Carol a lot, since “she was [her] only daughter.” The jury recommended the death penalty. The trial court agreed and sentenced Wilson to death for aggravated murder (based on the evading-kidnapping aggravator) and to imprisonment for the kidnapping and aggravated arson. B.

Procedural History On October 12, 1994, the state court of appeals on direct appeal affirmed, Wilson’s convictions and death sentence. Wilson v. Mitchell, No. 1:99-cv-0007, slip. op. at 5 (N.D.Ohio Jan. 14, 2003) (discussing state appeals). On January 24, 1996, the Ohio Supreme Court affirmed. Id. at 13. On January 3, 1997, the trial court rejected Wilson’s petition for postconviction relief. Id. at 18. On June 24, 1998, the court of appeals affirmed. Id. at 20. On November 4, 1998, the Ohio Supreme Court dismissed the appeal as not involving any substantial constitutional question. Id. at 22.

After filing his petition for postconviction relief on September 20, 1996, Wilson filed a motion to reopen his appeal (Mur-nahan application) on December 12, 1996. Id. at 22. The court of appeals denied this application as untimely on January 19, 1997. On October 22, 1997, the Ohio Supreme Court affirmed. On July 2, 1999, Wilson filed a petition for habeas corpus relief in the district court. Id. at 27. On January 14, 2003, the district court denied relief. Id. at 121.

The district court also granted a certificate of appealability as to one of Wilson’s claims (Claim 10: improper burden shifting on voluntary-intoxication defense). Id. On February 12, 2003, the district court partially granted Wilson’s motion to alter or amend judgment and issued a certificate of appealability as to four additional claims. On April 4, 2005, we partially granted Wilson’s motion for a certifícate of appealability, granting the certificate as to a portion of one additional claim. Thus, Wilson was granted a certificate of appeal-ability as to six issues, and he now raises five of them. II.

DISCUSSION A. Standard of Review We review de novo a district court’s decision to grant or deny a petition for a writ of habeas corpus. Joseph v. Coyle, 469 F.3d 441, 449 (6th Cir.2006) (citing Burton v. Renico, 391 F.3d 764, 770 (6th Cir.2004)). Because Wilson filed his habeas petition after the effective date of the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), its provisions apply to his case. Id. (citing Woodford v. Garceau, 538 U.S. 202, 210, 123 S.Ct. 1398, 155 L.Ed.2d 363 (2003), and Lindh v. Murphy, 521 U.S. 320, 336, 117 S.Ct. 2059, 138 L.Ed.2d 481 (1997)).

Under AEDPA, a federal court may grant a writ of habeas corpus with respect to a “claim that was adjudicated on the merits in State court proceedings” if the state court’s decision “was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States.” 28 U.S.C. § 2254(d)(1). A habe-as petition may also be granted if the state court’s decision “was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d)(2).

A state-court decision is contrary to clearly established federal law “if the state court applies a rule that contradicts the governing law set forth in [the Supreme Court’s] cases” or “if the state court confronts a set of facts that are materially indistinguishable from a decision of [the Supreme] Court and nevertheless arrives at a result different from [that] precedent.” Williams v. Taylor, 529 U.S. 362, 405, 406, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000).

A state-court decision is an unreasonable application of clearly established federal law if it “correctly identifies the governing legal rule but applies it unreasonably to the facts of a particular prisoner’s case,” id. at 407-08, 120 S.Ct. 1495 or if it “either unreasonably extends or unreasonably refuses to extend a legal principle from Supreme Court precedent to a new context,” Seymour v. Walker, 224 F.3d 542, 549 (6th Cir.2000). B. Procedural Considerations A petitioner seeking a writ of habeas corpus must meet certain procedural requirements to permit federal review of his habeas claims. Smith v.

Ohio Dep’t of Rehab. & Corr., 463 F.3d 426, 431 (6th Cir.2006). The petitioner must first exhaust the remedies available in state court by fairly presenting his federal claims to the state courts; unexhausted claims will not be reviewed by a federal court. Id. (citing Deitz v. Money, 391 F.3d 804, 808 (6th Cir.2004), and Lott v. Coyle, 261 F.3d 594, 601 (6th Cir.2001)).

The exhaustion “requirement is satisfied when the highest court in the state in which the petitioner was convicted has been given a full and fair opportunity to rule on the petitioner’s claims.” Lott, 261 F.3d at 608 (internal quotation marks and citation omitted). A federal court will not review claims that were not entertained by the state court due to the petitioner’s failure to (1) raise those claims in the state courts while state remedies were available, or (2) comply with a state procedural rule that prevented the state courts from reaching the merits of the claims. Lundgren v.

Mitchell, 440 F.3d 754, 763 (6th Cir.2006). For noncompliance with a state procedure to serve as a bar to habeas review, the state procedure must satisfy the standards set forth in Maupin v. Smith, 785 F.2d 135, 138 (6th Cir.1986). Smith v. Ohio Dep’t, 463 F.3d at 431. First, there must be a state procedure in place that the petitioner failed to follow. Maupin, 785 F.2d at 138. Second, the state court must have actually denied consideration of the petitioner’s claim on the ground of the state procedural default. Id.

Third, the state procedural rule must be an “adequate and independent state ground” to preclude habeas review. Id. This inquiry “generally will involve an examination of the legitimate state interests behind the procedural rule in light of the federal interest in considering federal claims.” Id. A state procedural rule must be “ ‘firmly established and regularly followed’ ” to constitute an adequate basis for foreclosing habeas review. Deitz, 391 F.3d at 808 (quoting Ford v. Georgia, 498 U.S. 411, 423-24, 111 S.Ct. 850, 112 L.Ed.2d 935 (1991)).

A state procedural rule is an independent ground when it does not rely on federal law. Coleman v. Thompson, 501 U.S. 722, 732, 111 S.Ct. 2546, 115 L.Ed.2d 640 (1991). If these three factors are satisfied, the petitioner can overcome the procedural default by either “demonstratfing] cause for the default and actual prejudice as a result of the alleged violation of federal law, or demonstrating] that failure to consider the claims will result in a fundamental miscarriage of justice.” Id. at 750, 111 S.Ct. 2546. C.

Wilson’s Claims Wilson raises five claims: (1) the erroneous instruction regarding voluntary intoxication was not harmless error; (2) the prosecution violated Brady v. Maryland by failing to disclose a Youth Services report to Wilson’s expert, Dr. Eisenberg, until Dr. Eisenberg was on the stand; (3) a juror was improperly excluded; (4) the trial court improperly instructed the jury regarding Wilson’s unsworn statement; and

(5) Wilson’s appellate counsel was ineffective for failing to raise the Brady claim on direct appeal. We discuss each of these claims in turn. 1. Impermissible Burden Shifting on Voluntary-Intoxication Defense The capital specification used against Wilson in the penalty phase of his trial alleged that he killed to avoid apprehension or detection and future trial for the offense of kidnapping. The Ohio Supreme Court concluded that the trial court — in the guilt phase — improperly shifted the burden to Wilson to prove that he lacked the specific intent, due to intoxication, to commit kidnapping.

The Ohio Supreme Court held, however, that this error was harmless. The Ohio Supreme Court conducted harmless-error analysis with respect to the kidnapping charge only; it did not conduct any harmless-error review of the evading-kidnapping death specification, which incorporated the identical burden-shifting instruction of the kidnapping charge from the guilt phase. The district court also concluded, for different reasons, that the error was harmless. Wilson chai-lenges these rulings, and the State admits that Wilson properly preserved this claim for habeas review.

As mentioned, at the penalty phase of the trial, the State elected to proceed only on Count I (murder by prior calculation and design) and specification one (evading detection or punishment for kidnapping).

The trial court instructed the jury to refer to the definition of kidnapping in the instructions regarding the kidnapping count when determining whether this capital specification existed. (Joint Appendix (“JA”) 1268 (explaining that, as to specification one to count one, “[t]he offense of kidnapping is defined for you in Count Four [kidnapping] of this charge”).) Those instructions, in turn, included instructions regarding the voluntary-intoxication defense: For purposes of Count Four [kidnapping], and any count or specification where Kidnapping is an element, and only for Count Four, or any count or specification where Kidnapping is an element, you may consider the defense of voluntary intoxication.

Intoxication exists when a person consumes a quantity of intoxicating beverage containing alcohol sufficient to advers[e]ly affect his mental processes and to deprive him of that clearness of intellect that he would otherwise have possessed. Intoxication is not an excuse for an offense. However, such evidence is admissible for the purpose of showing that the Defendant was so intoxicated that he was incapable of having the knowledge to commit the offense of Kidnapping. Knowledge is the element of this offense; and intoxication, even severe intoxication[,] can co-exist with knowledge.

On this issue, the burden of proof is upon the Defendant to establish by a preponderance or greater weight of the evidence that at the time in question he was so influenced by alcohol that he was incapable of having the knowledge to commit the offense.

If you find by a preponderance or greater weight of the evidence that the Defendant was incapable of having the knowledge to commit the offense, then you must find that the Defendant was not guilty of the offense of Kidnapping because knowledge is an essential element of the offense as I have previously instructed you. (JA 1269-72 (emphasis added).) Wilson contends that this instruction was erroneous and that the error was not harmless. a.

Whether the Instruction Was Erroneous “[T]he Due Process Clause protects the accused against conviction except upon proof beyond a reasonable doubt of every fact necessary to constitute the crime with which he is charged.” In re Winship, 397 U.S. 358, 364, 90 S.Ct. 1068, 25 L.Ed.2d 368 (1970). Additionally, due process prohibits requiring an accused to disprove an element of the crime charged. See Mullaney v. Wilbur, 421 U.S. 684, 704, 95 S.Ct. 1881, 44 L.Ed.2d 508 (1975).

The Ohio Supreme Court concluded that the instruction here violated this prohibition by requiring Wilson to disprove knowledge, an element of the kidnapping offense. State v. Wilson, 659 N.E.2d at 306 (“This instruction is unconstitutional under Win-ship because it required Wilson to disprove ‘knowledge,’ which is an element of the offense of kidnapping.”). The Ohio Supreme Court also explained, however, that Wilson made no claim that any error in this regard affected the death sentence in his case. Id.

Yet the district court concluded that Wilson did claim that this same error affected the death sentence because it was incorporated in the evading-kidnapping specification. The district court further held that, though one could argue that the confusing instruction was not technically improper because the burden shifting occurred in the context of an affirmative defense, the Ohio Supreme Court properly determined that the instruction was unconstitutional (as to the kidnapping charge). Wilson, slip op. at 95 n. 69.

The district court accordingly concluded that the instruction was unconstitutional as to the evading-kidnapping specification. We are not certain that an error regarding the knowledge element of a kidnapping offense necessarily translates into an error regarding the knowledge element of an evading-kidnapping specification.

In other words, one might say it is conceivable that a person could lack the requisite knowledge to commit kidnapping, yet have the requisite knowledge to commit murder to evade detection for kidnapping — for example, where the person believes he has committed kidnapping (but actually has not, perhaps because of earlier intoxication), and then commits murder to evade detection for the kidnapping he (erroneously) believes took place. But we do not decide this question. Instead, we assume that the instruction was erroneous with regard to the evading-kidnapping specification and address whether it was harmless. b.

The Error Was Harmless In assessing whether this error was harmless, we first provide a brief overview of the Ohio Supreme Court’s and district court’s different analyses, then provide a discussion of the applicable law regarding harmless-error review in this context, and, finally, apply that law to Wilson’s case. i.

Ohio Supreme Court’s and District Court’s Harmless-Error Rulings The Ohio Supreme Court concluded that the burden-shifting error was harmless in the context of the kidnapping charge: “[W]e find the error to be harmless under the facts of this case since the kidnapping of Lutz continued into the late morning and early afternoon. At that point, he clearly knew what he was doing and intoxication would not reasonably be available as a defense to negate ‘knowledge.’ ” State v. Wilson, 659 N.E.2d at 306.

As mentioned, the Ohio Supreme Court did not undertake harmless-error analysis with regard to the evading-kidnapping death specification; the court concluded that Wilson made no argument to that specific point. The district court disagreed with the Ohio Supreme Court’s reasoning. The district court concluded that Wilson met the standard under Brecht v.

Abrahamson, 507 U.S. 619, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993), to show that this error affected the outcome of the trial: “Clearly, this trial error[,] which permitted the jury to find Wilson guilty of kidnapping and the single aggravating circumstance to Count One even if the State had not sustained its burden with respect to the knowledge element, had ‘a substantial and injurious effect or influence in determining the jury’s verdict.’ ” Id. (quoting Brecht, 507 U.S. at 637, 113 S.Ct. 1710).

The district court explained further: “Had the jury been more clearly instructed as regards the burden of proof, it is possible that it might have concluded, in the face of all of the testimony and Wilson’s assertion of intoxication, that the State could not prove beyond a reasonable doubt that Wilson had the requisite knowledge for either the kidnapping charge of the indictment or the kidnapping specification to the aggravated murder charge.” Id. at 96-97.

Therefore, had “the kidnapping specification been the only specification for which Wilson had been found guilty,” the district court stated that it “would probably be inclined to grant the writ on this claim because, finding a constitutional error, [the district court] would be left with the ‘grave doubt’ ” described in O’Neal v. McAninch, 513 U.S. 432, 435, 115 S.Ct. 992, 130 L.Ed.2d 947 (1995), about whether that error is harmless. Id. at 98. The district court ultimately denied relief, however, because it concluded that there was an independent reason to deem the error harmless.

The court explained that Wilson was found guilty of two capital specifications (which, as mentioned, establish eligibility for the death penalty) in addition to the evading-kidnapping specification: (1) committing the offense during kidnapping, and (2) committing the offense during an aggravated arson.

The district court further noted that, during the penalty phase of a trial, “where two or more aggravating circumstances arise from the same act or indivisible course of conduct and are thus duplicative, the duplicative aggravating circumstances will be merged for purposes of sentencing.” Id. at 99 (quoting State v. Jenkins, 15 Ohio St.3d 164, 473 N.E.2d 264 (1984) (¶ 5 of syllabus)).

The state trial court noted that, based on this law, the State viewed these two additional specifications as merging with the first specification (murder to evade kidnapping), and the State therefore chose to proceed on only the first specification. Id. at 99-100.

The district court concluded that “[h]ad the State proceeded instead on the third specification relating to aggravated arson, for which there is no challenge as to the jury instructions and no challenge as to the sufficiency of evidence, undoubtedly the jury would have recommended the death penalty.” Id. at 100. “Therefore,” the district court continued, “the result would have been no different, that is, there is no ‘actual prejudice.’ ” Id.

Concluding that the error regarding the burden-shifting with respect to the first specification was harmless in this way, the district court denied habeas relief on this claim. Id. ii. Applicable Law Regarding Harmless-Error Review of Capital-Sentencing Error To assess properly the Ohio Supreme Court and district court’s analyses here, we must consider the proper standard for applying harmless-error review, the doctrine’s emphasis on the error’s actual (not hypothetical) impact, and the doctrine’s application where the error occurs in the unique context of capital sentencing.

This subsection addresses those points. (

a) Overview of Harmless-Error Standard Before the enactment of AEDPA, the Supreme Court articulated two harmless-error standards. Eddleman v. McKee, 471 F.3d 576, 582 (6th Cir.2006). On direct review, “before a federal constitutional error can be held harmless, the court must be able to declare a belief that it was harmless beyond a reasonable doubt.” Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967).

On collateral review, however, the State’s burden is lessened: In those proceedings, courts should deem an error harmless unless the error “had substantial and injurious effect or influence in determining the jury’s verdict.” Brecht, 507 U.S. at 637, 113 S.Ct. 1710. “When Congress enacted AEDPA, it complicated this dichotomy” because AED-PA provides that habeas relief shall not be granted unless the state-court decision was either (1) “contrary to,” or involved an “unreasonable application of,” clearly established federal law as determined by the Supreme Court; or (2) based on an “unreasonable determination of the facts.” Eddleman, 471 F.3d at 582 (citing 28 U.S.C. § 2254(d)).

We nonetheless continued to apply only the Brecht “substantial- and-injurious-effect” standard after AED-PA’s enactment because we concluded that if a petitioner meets that standard, “he will surely have demonstrated that the state court’s finding that the error was harmless beyond a reasonable doubt — the Chapman standard — resulted from ... an unreasonable application of Chapman.” Id. (quoting Nevers v. Killinger, 169 F.3d 352, 355 (6th Cir.1999)). In light of the Supreme Court’s decision in Mitchell v.

Esparza, 540 U.S. 12, 124 S.Ct. 7, 157 L.Ed.2d 263 (2003), however, we reconsidered this position in' Eddleman and held that “AEDPA replaced the Brecht standard with the standard of Chapman plus AEDPA deference when, as here, a state court made a harmless-error determination.” Eddle-man, 471 F.3d at 583. In other words, when assessing a state court’s harmless-error review, we asked whether that review was “contrary to,” or an “unreasonable application of,” Chapman.

See id. at 585 (“We now must determine whether the [state-court] decision that admitting Ed-dleman’s confession was harmless error was contrary to, or an unreasonable application of, Chapman.”). While Wilson’s appeal was pending, the United States Supreme Court rejected this approach. In Fry v.

Pliler, — U.S. -, 127 S.Ct. 2321, 168 L.Ed.2d 16 (2007), the Court unanimously concluded that regardless whether a state court applied Chapman’s harmless-error standard on direct review (i.e., that the state must prove that the error was harmless beyond a reasonable doubt), a federal habeas court applies the stricter (more state-deferential) Brecht standard (i.e., harmless unless the error had substantial and injurious effect on the outcome). In so holding, the Court explained that AEDPA did not replace the Brecht standard. Id. at 2326-27.

The petitioner in Fry argued (just as the Ed-dleman court concluded) that, because of AEDPA, a federal habeas court conducting harmless-error review had to ask whether the state court “unreasonably applied” Chapman to determine whether habeas relief was warranted. Id.

The Supreme Court explained, however, that “it is implausible that, without saying so, AEDPA replaced the Brecht standard of ‘actual prejudice,’ with the more liberal AED-PA/Chapman standard which requires only that the state court’s harmless-beyond-a-reasonable-doubt determination be unreasonable.” Id. at 2327 (citations and internal quotation marks omitted). “That said,” the Court continued, “it certainly makes no sense to require formal application of both tests (AEDPA/Chapman and Brecht) when the latter obviously subsumes the former.” Id.

In other words, a federal habeas court technically applies Brecht in light of AEDPA, but because the Brecht test is stricter (i.e., tougher on the petitioner) than AEPPA/Chapman, any petitioner that meets the Brecht standard will necessarily meet the AEDBA/Chap-man standard. Thus, when conducting harmless-error review, we simply apply the Brecht standard and ask whether Wilson has shown that the error had substantial and injurious effect in determining the jury’s verdict. (

b) Harmless-Error Review Looks to Actual, Not Hypothetical, Impact Characterizing an error as harmless might have either of two meanings. On the one hand, an error might be deemed harmless if it played such an inconsequential role in the actual trial in which it occurred that it assuredly had no impact on the trial’s verdict. 2 R. Hertz & J. Liebman, Federal Habeas Corpus Practice & Procedure § 31.4d (5th ed.2005).

On the other hand, an error might be deemed harmless — even if it played an important role in the actual trial' — 'if a hypothetical new trial absent the error would likely produce the same outcome as did the actual trial. Id.

The Supreme Court has indicated that of these two meanings the proper one is the first (i.e., whether the error had an actual impact on the outcome), and not the second (i.e., whether a hypothetical new trial would likely produce the same result): Consistent with the jury-trial guarantee, the question ... the reviewing court [is] to consider is not what effect the constitutional error might generally be expected to have upon a reasonable jury, but rather what effect it had upon the guilty verdict in the case at hand.

Harmless-error review looks, we have said, to the basis on which “the jury actually rested its verdict.” The inquiry, in other words, is not whether, in a trial that occurred without the error, a guilty verdict would surely have been rendered, but whether the guilty verdict actually rendered in this trial was surely unattributable to the error. That must be so, because to hypothesize a guilty verdict that was never in fact rendered — no matter how inescapable the findings to support that verdict might be — would violate the jury-trial guarantee. Sullivan v.

Louisiana, 508 U.S. 275, 279, 113 S.Ct. 2078, 124 L.Ed.2d 182 (1993) (citations omitted) (quoting Yates v. Evatt, 500 U.S. 391, 404, 111 S.Ct. 1884, 114 L.Ed.2d 432 (1991)). Likewise, as the Brecht Court explained, “[t]he standard for determining whether habeas relief must be granted is whether ... the ... error ‘had substantial and injurious effect or influence in determining the jury’s verdict’ ” Brecht, 507 U.S. at 623, 113 S.Ct. 1710 (quoting Kotteakos v. United States, 328 U.S. 750, 776, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946) (emphasis added)). (

c) Harmless Error in Capital Sentencing Crucial to this appeal is how the harmless-error principles discussed above apply in the capital-sentencing context when, as here, the jury considers an invalid aggravating factor when imposing a death sentence. One question is whether federal habeas courts can even conduct harmless-error review in that situation. The Supreme Court’s recent opinion in Brown v. Sanders, 546 U.S. 212, 126 S.Ct. 884, 163 L.Ed.2d 723 (2006), casts some doubt on our current view that federal courts can do so.

To fully assess these issues, one must first consider the development of the law in this area, including the Supreme Court’s past reliance on the distinction between so-called “weighing States” and “non-weighing States.” Since Furman v. Georgia, 408 U.S. 238, 92 S.Ct. 2726, 33 L.Ed.2d 346 (1972) (per curiam), the Supreme Court has required States to limit the class of murderers to which the death penalty may be applied. Sanders, 546 U.S. at 216, 126 S.Ct. 884.

This narrowing requirement is usually met when the trier of fact finds at least one statutorily defined eligibility factor at either the guilt or penalty phase. Id. (citation omitted). Once the narrowing requirement has been satisfied, the sentencer is called upon to determine whether a defendant found eligible for the death penalty should receive it. Id. Most States channel this function by specifying the aggravating factors (sometimes identical to the eligibility factors) that are to be weighed against mitigating considerations. Id.

The question facing courts in cases like the present one is what happens when the sentencer imposes the death penalty after at least one valid eligibility factor has been found, but under a scheme in which an eligibility factor or a specified aggravating factor is later held to be invalid. Id. To answer that question, the Supreme Court has distinguished between so-called weighing and non-weighing States. Id.

This terminology is somewhat misleading because the Court has held that in all capital cases the sentencer must be allowed to weigh the facts and circumstances that arguably justify a death sentence against the defendant’s evidence. Id. at 217-18, 125 S.Ct. 2884 (citing Eddings v. Oklahoma, 455 U.S. 104, 110, 102 S.Ct. 869, 71 L.Ed.2d 1 (1982)). The Court identified as weighing States those in which the only aggravating-factors permitted to be considered by the sentencer were the specified eligibility factors. Id. (citations omitted). Ohio is such a weighing state. See, e.g., Lundgren, 440 F.3d at 770.

Because the eligibility factors by definition identify distinct and particular aggravating features, if one of them is invalid then the jury cannot consider the facts and circumstances relevant to that factor as aggravating in some other capacity. Sanders, 546 U.S. at 218, 126 S.Ct. 884. In a weighing State, therefore, the sentencer’s consideration of an invalid eligibility factor necessarily skews its balancing of aggravators with mitigators. Id. (citation omitted). By contrast, a non-weighing State permits the sentencer to consider aggravating factors different from, or in addition to, the eligibility factors.

Id. (It would be clearer to call these States “complete weighing States,” because the jury can weigh everything that is properly admissible. See id. at 229-30, 126 S.Ct. 884 (Stevens, J., dissenting)). Because the sen-tencer can consider aggravating factors that are different from the eligibility factors, an invalid eligibility factor does not automatically skew the sentence as it does in a weighing state. Id. at 217, 126 S.Ct. 884.

The question here is a reviewing court’s role when an invalid eligibility factor (i.e., evading kidnapping), in a weighing State like Ohio, skews the jury’s balance of mitigating circumstances against that aggravating factor. Supreme Court decisions provide some reason to believe that a federal habeas court is simply not permitted to conduct harmless — error review — only a state court can do so. In Stringer v.

Black, for example, the Supreme Court explained that an invalid aggravating factor “in the weighing process invalidates the sentence and at the very least requires constitutional harmless-error analysis or reweighing in the state judicial system.” 503 U.S. 222, 237, 112 S.Ct. 1130, 117 L.Ed.2d 367 (1992) (emphasis added). Additionally, in Richmond v.

Lewis, the Court stated, “Where the death sentence has been infected by a vague or otherwise constitutionally invalid aggravating factor, the state appellate court or some other state sentencer must actually perform a new sentencing calculus.” 506 U.S. 40, 49, 113 S.Ct. 528, 121 L.Ed.2d 411 (1992) (emphasis added). We relied on these decisions when deciding cases involving invalid aggravating factors in weighing States, requiring States to conduct the new sentencing calculus. For example, in Houston v.

Dutton, 50 F.3d 381 (6th Cir.1995), a Tennessee (weighing State) jury sentenced the defendant after finding that the State established the “heinous, atrocious, or cruel” aggravator. Id. at 387. The State admitted on appeal that the aggravator was invalid because of an overly vague instruction. Id. Relying on Richmond’s language quoted above, we explained that habeas relief was properly granted because the Tennessee courts did not conclude that the instruction was erroneous and therefore had not performed “a new sentencing calculus.” Id. (emphasis added); accord Cone v.

Bell, 492 F.3d 743, 752 (6th Cir.2007) (“Cone is not entitled to a new sentence unless the Tennessee Supreme Court did not (1) conduct a proper harmless error analysis; or (2) reweigh the mitigating and aggravating factors in examining his sentence.” (citing Stringer, 503 U.S. at 230, 112 S.Ct. 1130)). In Coe v.

Bell, 161 F.3d 320 (6th Cir.1998), however, we held that, although we “may not perform reweighing” when a jury considers an invalid aggravator in a weighing state, we may “engage in harmless-error analysis.” Id. at 334. “In reweighing,” we explained, “a state court effectively vacates the original sentence and resentences the defendant; this process is hardly appropriate in the course of collateral review by a federal court.” Id. “In harmless-error analysis, by contrast, a court determines that the original sentence is not constitutionally infirm in the first place, a process that is quite appropriately performed on federal collateral review.” Id.

The Coe decision explained that in Houston we did not address the harmless-error question; rather, we held only that reweighing must be performed by a state court. Id. at 335. Further, the Coe decision explained that conducting harmless-error analysis as a federal habeas court was consistent with the Supreme Court’s statement in Richmond that state reweighing is required when “the death sentence has been infected by a constitutionally ... invalid aggravating factor” because, “by definition, ... an error that is harmless does not ‘infect’ the sentence and does not require reweighing by the state.” Id.

Finally, Coe reconciled Stringer’s language requiring “constitutional harmless-error analysis or reweighing in the state judicial system” by concluding that “the phrase ‘state judicial system’ modifies ‘reweighing’ only, and not ‘harmless-error analysis.’ ” Id. (emphasis added). Coe then concluded that the instructional error there — an overly vague instruction regarding the “heinous, atrocious, or cruel” aggravator — was harmless (under the Brecht standard) because the jury ignored the problematic aspect of the instruction. Id. at 336.

Coe’s holding — that a federal habeas court can conduct harmless-error review where a jury considers an invalid aggravator in a weighing State — continued as the law in this Circuit. See, e.g., Cone v. Bell, 359 F.3d 785, 798 (6th Cir.2004) (conducting such a harmless-error analysis after noting that Coe “drew a distinction between re-weighing and harmless error analysis and held that a federal habeas court is permitted to undertake the latter”), rev’d on other grounds by Bell v.

Cone, 543 U.S. 447, 459-60, 125 S.Ct. 847, 160 L.Ed.2d 881 (2005) (holding that we erred in concluding that state court failed to cure faulty “heinous, atrocious, or cruel” aggravator instruction); see also Jennings v. McDonough, 490 F.3d 1230, 1252 (7th Cir.2007) (noting that the Seventh Circuit had “yet to endorse federal harmless error review of death sentences based on invalid sentencing factors when the state appellate court has not performed its own harmless error analysis” and joining the “five circuit courts of appeals [that] have authorized such an approach”) (citing Coe, 161 F.3d 320).

Our holding in Coe is more questionable in light of the Supreme Court’s 2006 decision in Sanders. To be sure, the Sanders Court was faced with harmless-error in the context of a non-weighing State.

The Court explained that the “weighing/non-weighing scheme is accurate as far as it goes, but it now seems ... needlessly complex....” Sanders, 546 U.S at 219, 126 S.Ct. 884. “We think it will clarify the analysis,” the Court continued, “and simplify the sentence-invalidating factors we have hitherto applied to non-weighing States, if we are henceforth guided by the following rule: An invalidated sentencing factor (whether an eligibility factor or not) will render the sentence unconstitutional by reason of its adding an improper element to the aggravation scale in the weighing process unless one of the other sentencing factors enables the sentencer to give aggravating weight to the same facts and circumstances.” Id. at 220, 126 S.Ct. 884 (citation and footnote omitted) (first emphasis added).

In other words, “[i]f all the evidence was properly admitted and if the jury can use that evidence when it considers other aggravating factors, any error ... must be harmless.” Id. at 239, 126 S.Ct. 884 (Stevens, J., dissenting). This rule apparently modifies the analysis for non-weighing States, but leaves intact the Court’s prior jurisprudence regarding weighing states. See Hertz & Liebman, § 31.3 (6th ed.

Supp.2006) (noting that “the pre-Sanders jurisprudence for ‘weighing states’ ... apparently remains intact” but that Sanders “reshaped the analysis ... [it had] hitherto applied to non-weighing States”) (internal quotation marks and citations omitted); Adams v. Bradshaw, 484 F.Supp.2d 753, 787 n. 6 (N.D.Ohio2007) (noting that Sanders “does not apply” to invalid-aggravator claim under Ohio law because Sanders “involves a non-weighing state”).

When discussing weighing States, however, the Supreme Court in Sanders made a statement that might be taken to undercut Coe’s holding that a federal, not state, court may conduct harmless-error review where a jury considers an invalid aggravator. The Supreme Court first noted, as we did in Coe, that “[i]n a weighing State ... the sentencer’s consideration of an invalid eligibility factor necessarily skewed its balancing of aggravators with miti-gators.” Sanders, 546 U.S. at 217, 126 S.Ct. 884 (citing Stringer, 503 U.S. at 232, 112 S.Ct. 1130).

The Supreme Court then stated that, under Stringer, this skewing “required reversal of the sentence (unless a state appellate court determined the error was harmless or reweighed the mitigating evidence against the valid aggravating factors).” Id. (citing Stringer, 503 U.S. at 232, 112 S.Ct. 1130) (emphasis added). This reading of Stringer implicitly rejects the Coe Court’s

interpretation that Stringer’s language requiring “constitutional harmless-error analysis or reweighing in the state judicial system” allows a federal habeas court to conduct harmless-error review and merely limits reweighing to states. See Coe, 161 F.3d at 335 (noting that the phrase “state judicial system” in Stringer “modifies ‘reweighing’ only, and not ‘harmless-error analysis’ ”); cf.

Adams, 484 F.Supp.2d at 788(“Recently the Supreme Court [in Sanders ] noted that in a weighing state, the sentencer’s consideration of an invalid eligibility factor necessarily upsets its balancing of the aggravating circumstances with the mitigating factors requiring reversal of the sentence unless a state appellate court determined the error was harmless or reweighed the mitigating evidence against the valid aggravating factors.”) (citation omitted) (emphasis added).

Leading commentators appear to share this view: “[I]n a weighing State, when an eligibility or aggravating factor is found to have been invalid, the federal courts may not themselves engage in either a reweighing or in harmless error analysis; the condemned individual has a constitutional right to have either the state courts or the original sentencer reweigh the valid aggravating and mitigating factors.” Hertz & Liebman § 31.3 (6th ed. Supp.2006) (discussing Sanders and citing cases such as Richmond).

Although Sanders’s statements imply that only a state court may conduct harmless-error review in this situation, those statements are dicta, see Jennings, 490 F.3d at 1252 (noting that none of the Supreme Court decisions regarding this issue “squarely addresses the issue of federal district courts conducting harmless error review in place of state courts”), and do not demand that we change our current state of the law. Indeed, the Seventh Circuit’s recent endorsement of our view in Coe (that federal courts may conduct harmless-error review in this context) considered Sanders. See id.

In light of these considerations, we continue to hold that federal courts may conduct harmless-error review of invalid aggravating factors even where the state court has not done so. Though a contrary holding would be plausible in light of Sanders’s language, cf. Eddleman, 471 F.3d at 583 (“Today, we reconsider our position in light of the Supreme Court’s decision in Mitchell v.

Es-parza, which strongly implied that courts should apply only the Chapman plus AED-PA deference standard of review.” (emphasis added)), we believe that should arise only from a clear statement from our en banc court or the United States Supreme Court. iii. Application to Wilson’s Appeal Having concluded that the Ohio Supreme Court did not conduct a harmless-error review of the (presumed) invalid evading-kidnapping aggravating factor, and having concluded that we may nonetheless conduct harmless-error review, we determine that any improper-burden shifting in this regard was harmless.

The only element of the kidnapping charge and specification at issue is the knowledge element. The trial court instructed the jury that “[a] person acts knowingly, regardless of his purpose, when he is aware that his conduct will probably cause a certain result.

A person has knowledge of circumstances when he is aware that such circumstances probably exist.” (JA 1270-71.) The court continued, “Knowingly means that a person is aware of the existence of the facts and that his acts will probably cause a certain result.” (JA1271.) After these instructions, the trial court provided the voluntary-intoxication instructions referred to above.

To recap, those instructions provided that “[ijntoxi-cation exists when a person consumes a quantity of intoxicating beverage containing alcohol sufficient to advers[e]ly affect his mental processes and to deprive him of that clearness of intellect that he would otherwise have possessed.” (JA 1271.) The trial court further instructed that “[ijntoxieation is not an excuse for an offense,” but that “such evidence is admissible for the purpose of showing that the Defendant was so intoxicated that he was incapable of having the knowledge to commit the offense of Kidnapping.

Knowledge is the element of this offense; and intoxication, even severe intoxication[,] can co-exist with knowledge.” (JA 1271.) At this point, the trial court provided the (presumed) improper burden-shifting instruction (placing the burden on Wilson to establish that he was so intoxicated that he was incapable of having the knowledge to commit the offense). (JA 1272.) The district court explained that the record reveals the following testimony with respect to Wilson’s drinking: 1. Bonnie Menges, who managed and bartended at the Empire Tavern, testified that on May 3, 1991, she worked from 3:00 p.m. to 7:00 p.m.

When she arrived shortly before her shift, Wilson was already there, and when she left he was still there. He was drinking vodka and orange juice. Although Menges did not believe Wilson was intoxicated, he “was putting them down pretty good that day[,J” “faster than he normally does.” Wilson, slip. op. at 96 n. 71. 2. Doug Pritt, Lutz’s good friend whose band played at the Empire Tavern, testified that he arrived there around 9:30 p.m. on May 3, 1991. According to Pritt, Wilson did not arrive until around 10:30 p.m.

Pritt left the bar at 1:00 a.m. on May 4,1991; Lutz walked out with him to say goodbye and then returned to the bar. Pritt testified that Lutz was drinking beer and he saw her offer to buy Wilson a drink, but he did not know what kind. Id. 3. Gregory McKinney, who worked at the Empire Tavern and had the 7:00 p.m. to 3:00 a.m. shift on May 3-4, 1991, testified that he was not sure when Wilson came into the Tavern, but estimated it was about 9:00 or 10:00 p.m. On cross-examination, McKinney admitted that Wilson could have been there when he arrived for his shift, but was not sure.

Wilson was drinking beer and vodka and orange juice and shooting pool with Lutz and her friend, Doug Pritt. After Pritt left at about 12:30 a.m., Lutz remained. Although Lutz and Wilson were still drinking beer and, in addition, “had a few shots of Jack Daniels,” McKinney did not believe they were drunk. Id. 4. Darlene DeBolt, Wilson’s good friend, testified that she saw him a few minutes before 6:00 a.m. on May 4, 1991, as she was opening the Gas-town business where she worked. She “asked him if he had been drinking all night because [she] smelled alcoholic beverages on him.” He said that he had a few beers.

He did not appear intoxicated to her. Id. at 96-97 n. 71. 5. Rodney “Lee” Mele, Wilson’s cousin, testified that he first saw Wilson around 9:30 or 10:00 p.m. on May 3, 1991 “sitting on his car out in the street.” Around 10:30 p.m., the two of them walked to a local Convenient store where Wilson bought a six-pack of beer. Id. at 97 n. 71. 6. Additionally, there were apparently tapes played for the jury of the interview conducted by Detective Riley in which Wilson stated that he had spent approximately $25 to $30 on alcohol on the afternoon of May 3, 1991, and the same amount in the evening.

He further stated that he and Lutz had consumed more beer at his trailer after they left the bar when it closed in the early morning hours of May 4,1991. Id. During the penalty phase, defense expert Dr.

Robert Forney, a forensic toxicologist, testified on cross-examination that, although an alcoholic drink that a person had at 2:30 a.m. (the approximate time that Wilson left the bar with Lutz) would be completely gone from his or her system by 1:30 p.m. (the time the fire department was called to the scene of the car fire), if that person were an alcoholic, or a heavy chronic drinker, “there is residual damage to the brain and nervous system which may remain even when the alcohol is gone.” Id. at 97 n. 72.

Therefore, it was “not [his] opinion that judgment ... would be unaffected or that the brain would be unaffected during periods when alcohol was absent....” Id. Evidence suggested that Wilson was an alcoholic. Id. (noting, for example, testimony that Wilson “never went a single day without drinking”). Countering this general evidence was specific, and strong, evidence that Wilson had the requisite knowledge for the evading kidnapping-aggravator, that is, that he committed the murder to evade detection for kidnapping Lutz.

The Ohio Supreme Court referred to portions of Wilson’s statement to the police in which he stated that he recalled thinking, when Lutz was locked in the trunk, “How am I going to get out of this?” State v. Wilson, 659 N.E.2d at 292 (syllabus). In this statement, Wilson also explained that after he let Lutz out to go to the bathroom, she said that “she’d go home and forget about it,” but that Wilson “didn’t believe her and thought to himself, ‘How can you forget about being locked in a trunk?’ ” Id.

When asked why he did not simply leave her in the trunk, Wilson replied that he “figured ... eventually somebody would find her. She’d get out and tell who [he] was.” Id. (emphasis added). The district court concluded that the error was not harmless.

The district court explained that the Ohio Supreme Court “made no factual findings” and “merely made the conclusory statement that ... the error was ‘harmless under the facts of this case [in the context of the kidnapping charge], since the kidnapping of Lutz continued into the late morning and early afternoon [when Wilson] ... clearly knew what he was doing....” Wilson, slip op. at 90 (quoting Wilson, 659 N.E.2d at 306). “Had the jury been more clearly instructed as regards the burden of proof,” the district court explained, “it is possible that it might have concluded, in the face of all of the testimony and Wilson’s assertion of intoxication, that the State could not prove beyond a reasonable doubt that Wilson had the requisite knowledge for either the kidnapping charge of the indictment or the kidnapping specification to the aggravated murder charge.” Id. at 97.

Although “it is possible” that the jury might not have found the evading-kidnapping specification beyond a reasonable doubt, we cannot say, as we must to grant relief under Brecht, that the burden-shifting error had substantial and injurious effect on the verdict. Crucial to our conclusion are Wilson’s own statements.

Although the Ohio Supreme Court did not reference those statements in the specific portion of its opinion addressing the burden-shifting error on the kidnapping charge, the court nonetheless relied on them, explaining, for example, that Wilson did not simply leave Lutz in the trunk of his car because “[s]he’d get out and tell who [he] was.” State v. Wilson, 659 N.E.2d at 292 (syllabus); cf. Cone, 492 F.3d at 752 (assessing harmless-error review and relying on Tennessee Supreme Court’s discussion of penalty-phase evidence “in a

section immediately preceding a discussion of the claims ... raised on appeal”). We cannot discern what Wilson’s statements such as this one could mean, if not that Wilson knew his actions could lead to kidnapping charges and he therefore committed the crime to escape detection.

Moreover, after receiving this erroneous instruction, the jury was properly instructed that the State must prove, beyond a reasonable doubt, “all of the essential elements of the offense of Kidnapping” to reach a guilty verdict on that count. (JA 1273.) In sum, the error did not “ha[ve] substantial and injurious effect or influence in determining the jury’s verdict.” Brecht, 507 U.S. at 623, 113 S.Ct. 1710 (citation and internal quotation marks omitted). Wilson is therefore not entitled to habeas relief on this claim.

Before addressing Wilson’s remaining claims, we note briefly a concern with the district court’s analysis. As mentioned, the district court concluded that, although the Ohio Supreme Court’s harmless-error analysis regarding the burden shifting was incorrect, the error at issue was harmless for an independent reason: Had the prosecution gone forward with the arson aggra-vator at sentencing, the district court said, the jury surely would have relied on that aggravator to impose death.

This analysis appears to posit improperly a hypothetical scenario forbidden by Supreme Court precedent when determining whether a state court properly deemed an error harmless. As discussed above, Sullivan explains that “to hypothesize a guilty verdict that was never in fact rendered — no matter how inescapable the findings to support that verdict might be— would violate the jury-trial guarantee”. Sullivan, 508 U.S. at 279, 113 S.Ct. 2078.

Similarly, for the district court to hypothesize a sentencing outcome, based solely on the arson aggravator, that was never in fact rendered — no matter how inescapable the findings to support that sentence might be — would seem to violate the jury-trial guarantee. “The question is ... not [whether the jurors] were ... right in their judgment, regardless of the error or its effect upon the verdict [or sentence].” Kotteakos, 328 U.S. at 764, 66 S.Ct. 1239. “It is rather what effect the error had or reasonably may be taken to have had upon the jury’s decision.” Id.

Simply put, “harmless-error review looks ... to the basis on which ‘the jury actually rested its verdict,’ ” Sullivan, 508 U.S. at 279, 113 S.Ct. 2078 (citation omitted), and Wilson’s jury did not actually rest his death sentence on the arson aggravator-they never even considered it. This is true even though the arson specification and the evading-kidnapping specification involve substantially the same course of conduct and therefore merged at sentencing. The jury indeed concluded— at the eligibility phase — that Wilson committed the killing during an aggravated arson.

But the prosecution chose to proceed at sentencing with the evading-kidnapping specification, and that is the sole aggravator the jury placed on death’s side of the scale. The jury never considered how the arson aggravator would tip that same balance. To hypothesize about that — “no matter how inescapable the findings to support that [death sentence] might be” — appears to improperly remove the question from the jury. See Sullivan, 508 U.S. at 279, 113 S.Ct. 2078. 2. Failure to Disclose Youth Services Report Under Brady v.

Maryland Wilson next contends that the State failed to disclose Wilson’s records from the Ohio Department of Youth Services (“DYS”), in violation of Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963). Wilson explains that he requested the DYS records before trial and that the prosecution denied ever having them. During the mitigation phase of the trial, however, the prosecutor used the records — in particular, a psychological report by DYS employee Hugh Turner — -to impeach Dr. Eisenberg’s testimony.

Wilson argues that the records were material because they were critical for the preparation of Wilson’s mitigation phase presentation. In particular, Wilson states that the records were necessary for use by Dr. Eisenberg in his preparation of Wilson’s psychological evaluation and mental-health assessment, and that, without the records, Dr. Eisenberg’s expert testimony was unprepared and incompetent. This claim, although not raised on direct appeal, is nonetheless preserved because the procedural rule upon which the state court relied was not firmly established and regularly followed.

Wilson correctly notes that the Ohio Supreme Court repeatedly ignored the ninety-day time limit under Ohio Rule of Appellate Procedure 26(

B) required for filing a Mumahan application, in which he raised this claim. In Franklin v. Anderson, 434 F.3d 412 (6th Cir.2006), we concluded that Rule 26(

B) did not satisfy the Maupin test because (1) the rule was not “firmly established and regularly followed” and therefore it was not an “adequate and independent state ground for foreclosing review” (the third prong of the Maupin test), and (2) the rule was not consistently enforced by the Ohio Supreme Court (the second prong of the Maupin test). Id. at 420 (internal quotation marks omitted). Accordingly, Wilson’s claim is not procedurally defaulted.

In Brady, the Supreme Court held that “the suppression by the prosecution of evidence favorable to an accused upon request violates due process where the evidence is material either to guilt or to punishment, irrespective of the good faith or bad faith of the prosecution.” 373 U.S. at 87, 83 S.Ct. 1194. The duty to disclose Brady evidence encompasses impeachment evidence as well as exculpatory evidence. United States v. Bagley, 473 U.S. 667, 676, 105 S.Ct. 3375, 87 L.Ed.2d 481 (1985); Norris v. Schotten, 146 F.3d 314, 334 (6th Cir.1998). We explained in United States v.

Bencs that “[m]ateriality pertains to the issue of guilt or innocence, and not to the defendant’s ability to prepare for trial.” 28 F.3d 555, 560 (6th Cir.1994) (citing United States v. Agurs, 427 U.S. 97, 112 n. 20, 96 S.Ct. 2392, 49 L.Ed.2d 342 (1976)). In particular, “[e]vi-dence is material ‘only if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different.’ ” United States v. Phillip, 948 F.2d 241, 249 (6th Cir.1991) (quoting Pennsylvania v. Ritchie, 480 U.S. 39, 57, 107 S.Ct. 989, 94 L.Ed.2d 40 (1987)).

Further, “Brady generally does not apply to delayed disclosure of exculpatory information, but only to complete failure to disclose.” Id. Lastly, “[d]elay ... violates Brady [only] when the delay itself causes prejudice.” United States v. Patrick, 965 F.2d 1390, 1400 (6th Cir.1992). During cross-examination, Dr. Ei-senberg testified first that he had an opportunity to review the Turner report, and second that “the report is consistent with what [he] ha[d] been saying....” (JA 1347.) The report was only two pages long, and Dr.

Eisenberg did not indicate that he needed more time, apart from the time he was given by the court, to review it. Further, Dr. Eisenberg stated that he generally agreed with the conclusions in the report. (JA 1347 (“I think the report is consistent with what I have been saying.”); JA 1348 (stating that he agreed with the report’s conclusion that Wilson had a “narcissistic frame of reference”); JA 1350-51 (agreeing with the report’s observation about Wilson’s lack of trust and agreeing with the report’s conclusion that Wilson’s interpersonal conflicts could lead to hostile, aggressive behavior).) Lastly, Dr.

Eisenberg testified, in an attempt to discredit the report, that Turner was not licensed as a psychologist at that time. Given that Dr. Eisenberg’s testimony indicated that he agreed -with much of the Turner report, Wilson has not established how Dr. Eisenberg’s testimony would have been any different if the report had been disclosed earlier, or how earlier disclosure of the report would have altered the outcome of the penalty phase.

Thus, although Wilson may be able to establish that the records were suppressed by the State despite Wilson’s repeated requests to obtain the records, Wilson has not demonstrated that even if the State had timely disclosed the records that there is a reasonable probability that the outcome of the proceeding would have been different. Accordingly, we conclude that Wilson’s Brady claim lacks merit and therefore does not warrant habeas relief. 3.

Improper Juror Dismissal Wilson next argues that the trial judge improperly dismissed a potential juror for cause because her statements regarding her inability to impose a death sentence were equivocal. The State concedes that Wilson has preserved this argument. He contends that O.R.C. § 2945.25(

C) incorporates the standard set forth in Witherspoon v. Illinois, 391 U.S. 510, 88 S.Ct. 1770, 20 L.Ed.2d 776 (1968), for the dismissal of a juror unable to impose a death sentence. Wilson claims that by applying the standard announced in Wainwright v. Witt, 469 U.S. 412, 105 S.Ct. 844, 83 L.Ed.2d 841 (1985), the Ohio Supreme Court erred in two ways: (1) applying the more lenient Witt standard violated Wilson’s liberty interest in the stricter With-erspoon standard, and (2) by applying the Witt standard the Ohio Supreme Court violated the Separation of Powers doctrine by overruling the Ohio legislature.

In Witt, the Supreme Court explained that a prospective juror may be excluded for cause because of his or her views on capital punishment when “the juror’s views would prevent or substantially impair the performance of his duties as a juror in accordance with his instructions and his oath.” 469 U.S. at 424, 105 S.Ct. 844 (internal quotation marks omitted). The Court elaborated that “this standard ... does not require that a juror’s bias be proved with unmistakable clarity.” Id. (internal quotation marks omitted).

Further, the Court noted that “deference must be paid to the trial judge who sees and hears the juror.” Id. at 426, 105 S.Ct. 844. The Court in Witt made clear, however, that it was merely clarifying its previous Wither-spoon decision. Id. at 424, 105 S.Ct. 844. (“We therefore take this opportunity to clarify our decision in Witherspoon .... ”).

Moreover, we have explained that a petitioner must show that the selected jury was biased to succeed on this claim: When reviewing a trial court’s dismissal of potential jurors for cause, this court must determine whether the trial court’s decision prevented the empaneling of an impartial jury. It is not enough for the defendant to show that the decision to exclude the two jurors was improper. He also must show that the jury selected was biased. Hill v. Brigano, 199 F.3d 833, 844 (6th Cir.1999). Wilson has made no claim that the jury empaneled was biased.

Without a showing that his jury was biased, merely arguing that the trial court erred in dismissing a prospective juror for cause is insufficient to warrant habeas relief. Hill, 199 F.3d at 844-45. This claim is without merit. 4. Erroneous Instruction Regarding Wilson’s Unsworn Statement Wilson next contends that the trial court improperly instructed the jury during the penalty phase that his unsworn statement was not evidence, and that this limited the jury’s consideration of relevant mitigating evidence. Wilson’s unsworn statement included evidence of his youth, his alcoholism, and his remorse for the offense.

The trial court instructed the jury as follows regarding the statement: In this phase, the Defendant made a statement, but he did not testify under oath and was not subject to cross-examination.

It is his right under Ohio law to make such a statement and this statement of the Defendant, although not considered evidence, may be considered by you for whatever purpose you would assign. (JA 1412-13 (emphasis added).) Assuming that this argument is properly preserved, it is without merit. “Under the Eighth Amendment, the jury in a capital case may ‘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.’ ” Mason v.

Mitchell, 320 F.3d 604, 618 (6th Cir.2003) (quoting Lockett v. Ohio, 438 U.S. 586, 604, 98 S.Ct. 2954, 57 L.Ed.2d 973 (1978)). “[A] jury instruction violates Lockett when there is a reasonable likelihood that the jury has applied the challenged instruction in a way that prevents the consideration of constitutionally relevant evidence.” Gall v. Parker, 231 F.3d 265, 324 (6th Cir.2000) (citing Boyde v. California, 494 U.S. 370, 380, 110 S.Ct. 1190, 108 L.Ed.2d 316 (1990)).

Wilson is entitled to habeas relief on this claim if the instruction “by itself so infected the entire trial that the resulting conviction violates due process.” Estelle v. McGuire, 502 U.S. 62, 72, 112 S.Ct. 475, 116 L.Ed.2d 385 (1991) (citation omitted). Although the trial court instructed that Wilson’s statement was not evidence, the court also instructed that the jury may consider it for whatever purpose it would assign. In short, there is no reasonable likelihood that the instructions prevented the jury’s consideration of the statement.

Additionally, even assuming that the instructions were problematic, the jury considered similar mitigating evidence from other witnesses at sentencing, so the instructions regarding his statement could not have so infected the entire trial as to violate due process. This claim is therefore without merit. 5. Ineffective Assistance of Appellate Counsel for Failure to Raise Brady Claim Wilson contends that appellate counsel were ineffective on direct appeal for failure to raise the Brady claim discussed above.

Because the Brady claim is without merit, Wilson cannot demonstrate that appellate counsel’s failure to raise this non-meritorious claim constituted ineffective assistance under Strickland v. Washington, 466 U.S. 668, 687-88, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). III. CONCLUSION For the foregoing reasons, we AFFIRM the denial of Wilson’s petition for a writ of habeas corpus. CONCURRENCE . Unless otherwise noted, these facts are from the Ohio Supreme Court's decision on direct appeal, State v. Wilson, 74 Ohio St.3d 381, 659 N.E.2d 292 (1996). .

The Ohio Supreme Court refers to the death “specifications” when discussing both (1) eligibility factors (the factors a jury must find to make the defendant eligible for the death sentence); and (2) aggravating factors (the factors the jury must weigh during the penalty phase to determine whether the defendant should actually receive the death sentence). .

Judge Rogers's concurring opinion contends that resolution of this issue is unnecessary because Coe and Sanders address harmless-error analysis in the context of remaining, valid aggravating factors — something not applicable to this case, which involves the harmlessness of a single aggravating factor. We are still assessing, however, whether the error embedded in that single aggravating factor was harmless in the context of a jury's capital-sentencing decision, which requires balancing that aggravating factor against mitigating circumstances — and this invalid aggravating factor skews that balance.

See Sanders, 546 U.S. at 217, 126 S.Ct. 884 ("In a weighing State ... the sentenced s consideration of an invalid eligibility factor necessarily skewed its balancing of aggrava-tors with mitigators.”) (emphasis added). We believe cases such as Coe and Sanders inform whether that skewing is harmless, even where there are no other aggravating factors. See, e.g., Coe, 161 F.3d at 336 (holding that instructional error regarding a single aggravating factor was harmless— not because other aggravating factors remained (though they did) — but because the jury ignored the problematic aspect of the instruction).

Moreover, although we agree with the concurrence that the Ohio Supreme Court likely would apply the same harmless-error analysis from the kidnapping-charge context to the evading-kidnapping aggravator, we are reluctant to uphold a state court’s harmless-error review of capital sentencing factors that did not actually occur.