Rovner, Circuit Judge.

Michael Ausbrook, Attorney, Bloomington, IN, for Petitioner-Appellee.

Jesse Drum, Attorney, Office of the Attorney General, Indianapolis, IN, for Respondent-Appellant.

Before Rovner, Hamilton, and Kirsch, Circuit Judges.

Rovner, Circuit Judge.

Dewayne Dunn was convicted in Indiana state court for the murder of Angel Torres and sentenced to a term of imprisonment of 58 years. The state's case against Dunn was based largely on the testimony of two pathologists. In a state court post-conviction proceeding, Dunn sought a new trial, arguing that his trial counsel was ineffective for failing to consult with any forensic pathologist. The court denied that relief, and the Indiana Court of Appeals affirmed the post-conviction court's decision. Dunn subsequently filed a petition for a writ of habeas corpus under 28 U.S.C. § 2254 based on the ineffective assistance of trial counsel, and the district court granted a conditional writ of habeas corpus. The state now appeals that grant of the writ to this court.

The district court, in assessing the merits of the pro se habeas petition, properly limited its consideration to only the facts as presented to the state court and did not conduct its own evidentiary hearing. See Shinn v. Martinez Ramirez , ––– U.S. ––––, 142 S. Ct. 1718, 1739–40, 212 L.Ed.2d 713 (2022). As did the district court, we recite the facts as set forth in the state court record and decision, which are not challenged by the parties.

I.

The incident that formed the basis of the criminal conviction occurred in Elkhart, Indiana, outside the apartment building in which Dunn resided. At the time, Dunn lived in a second-floor apartment with his former girlfriend, Letha Sims, and two of her sons. That apartment had a balcony with an external staircase to the ground that was shared with the adjacent apartment in which Angel Torres, the victim, resided. Although Sims and Dunn were still living together, according to Letha by the date of the incident they were no longer on good terms and were no longer intimate. Sims and Torres had sex on one occasion, but Sims described her relationship with him as one of close friends. Dunn and Torres were friends, and would frequently sit on the porch socializing, smoking and drinking.

On September 3, 2008, Damen Collins was riding his bike near the apartment building when he saw an altercation between persons later identified as Dunn and Sims taking place on a balcony, heard Sims screaming for help, and saw her being pushed halfway down the stairs. He saw another man, identified as Torres, come out of his apartment and Dunn pushed him back into his apartment. Collins was about 100 feet away from the balcony when he saw the altercation, and nothing was blocking his view. Collins then called 911 and left the area. Although Collins claimed that it was dusk at the time of the incident, that assertion is clearly wrong because the 911 call was received at 11:40 p.m.

Officers arrived on the scene about eight minutes after receiving the dispatch and found Torres lying at the bottom of the staircase in a pool of blood, unresponsive and with labored breathing. A baseball bat was underneath his body. Dunn was nearby along with Sims's son Jamar "Willie" Sims. Dunn was very agitated and shouting that he "didn't do anything."

Torres subsequently died as a result of his injuries, and Dunn was convicted of first degree murder. The evidence at trial consisted of expert testimony from the state by a forensic pathologist and a blood spatter expert, and, for the defense, eyewitness testimony from Sims and her son Willie, a high school junior.

Willie testified that he witnessed the altercation between Dunn and Torres that night, and recounted it as follows. Earlier in the day, several people including Sims, Torres, Dunn and other friends were all on the balcony drinking. Willie went to take a nap and awakened to hear Dunn screaming and hollering. Willie followed Dunn upstairs and while Willie was getting a drink of water, Dunn walked outside and Willie heard an argument ensue between Dunn and Torres, and heard Dunn say, "don't hit me with that bat." Willie then looked outside and observed Torres and Dunn fighting and saw Torres strike Dunn on the shoulder with a baseball bat. Willie saw the two wrestling over the bat after Torres struck Dunn with it. Both were perilously close to the stairs, with Dunn facing the stairs and Torres with his back to the stairs. As they struggled over the bat, Willie saw Torres fall backwards down the steps, landing on the banister first and then flipping over it and hitting the pavement below head first. Willie ran down the stairs to help and saw Torres in a pool of blood. He then ran to flag down a police officer. He never saw Dunn strike Torres, either with the bat or anything else.

Sims testified as well regarding the events of that day. She stated that earlier that day, she went next door to Torres's apartment to have a beer while Dunn took the dog for a walk. One beer led to more and to some Bacardi Rum, and when he returned from the walk Dunn joined them. They were all laughing and having a good time until an argument between Torres and Dunn broke out. Sims could not recall what prompted the argument, but said that Torres and Dunn argued a lot, especially when they were drinking. The argument spilled out onto the porch, and Sims and her other son went outside to break it up. She then went back inside Torres's apartment, and Torres and Dunn each went back into their apartments. At some time later, Dunn kicked the door to get back into Torres's apartment. Torres then grabbed the baseball bat and Torres and Dunn went back outside. Sims could hear the commotion outside through the screen door, and heard Dunn tell Torres to "stop hitting me with the bat." She went out onto the porch when she heard a thud, looked down, and saw Torres at the bottom of the stairs with the bat under him. When Sims came out of the apartment, she saw Dunn standing at the top of the stairs and observed him descend the stairs and begin tugging on Torres's arm, imploring him to get up.

The first officer arrived on the scene eight minutes after receiving the dispatch and observed Dunn standing next to Torres's body exclaiming "I didn't do anything!" Dunn told the officer that Torres had struck him with the bat and that a scuffle had ensued. The officer stated that Torres was lying on the baseball bat when he arrived, and that Dunn was upset but not acting aggressively. Another officer interviewed Sims, and her statements to him were consistent with her subsequent trial testimony, including telling the officer that she heard Dunn tell Torres to stop hitting him with the bat, and that she heard a thud, went outside, and saw Torres on the pavement below.

Other evidence corroborated at least some of that witness testimony. For instance, Dunn was taken to the hospital that evening, and photos of his back show abrasions on his shoulders consistent with being struck by a bat. In addition, when asked at the hospital how he got those injuries, Dunn told the evidence technician that he was struck by a baseball bat. He told the responding officer the same thing. The testimony as to the drinking that occurred that day was also corroborated by hospital records. According to the emergency room doctor who worked on Torres later that evening, Torres's blood alcohol content was .294, establishing that he was significantly under the influence of alcohol. Finally, in addition to the evidence that the bat was under Torres's body on the ground, forensic examination of the bat determined that Torres's blood was found on the handle and middle portion of the bat but not the barrel of the bat, which was consistent with the claims that the bat was used to strike Dunn but not Torres.

Torres died from his injuries, and an autopsy was conducted by Dr. Chrenka, who is board certified as an anatomical, clinical, and pediatric pathologist, but is not a board-certified forensic pathologist. Although he occasionally does autopsies, most of Dr. Chrenka's work does not involve doing them. He identified the cause of death as massive blunt trauma to the head. The determination of the manner of death, such as whether it could be deemed a homicide, suicide, or accident, was deemed to be "uncertain."

Nearly two years after Torres's death, the state charged Dunn with murder in his death. The state's theory was that Torres's death resulted from a beating that occurred after the fall, and that some object was used to beat him but not necessarily the bat. In fact, at closing argument, the state conceded that the evidence indicated the bat was not the weapon used in the beating.

At trial, the state presented expert testimony from a blood spatter expert and a forensic pathologist. Dr. Scott Wagner was a board-certified forensic pathologist who had testified on cases hundreds of times, 98% of them for the state. He was retained by the state to review the autopsy of Dr. Chrenka. Dr. Wagner testified that Torres had multiple skull fractures, with part of his skull "crushed like gravel" and his ribs broken as well as his clavicle, and that he experienced severe trauma. On cross-examination, defense counsel merely confirmed that Dr. Wagner was being called for his second opinion on the cause and manner of death. On redirect examination, Dr. Wagner testified that Dunn's injuries were not consistent with someone merely falling down the stairs.

Dean Marks, a sergeant of the Indiana State Police, who had specialized training in blood spatter analysis, testified as to that evidence. He testified that the pattern of Torres's blood that was found on the wall at the bottom of the staircase was caused by "impact spatter," which means that some degree of force was applied to a blood source, causing the blood to disperse. He maintained that the blood spatter pattern was not consistent with someone walking or even stomping through the blood that was pooled on the pavement, or with Torres simply falling down the stairs. Torres's blood was also spattered on a vehicle parked adjacent to the stairs and parallel to the bloodstained wall, and there was "cast-off blood staining" on the concrete between the staircase and the vehicle, indicating that a bloody instrument of some kind had been "swung" or flung."

In addition to that testimony, evidence was presented that Dunn's shoeprint was found on the outside of Torres's apartment door and the doorjamb was freshly splintered, and that blood was cast onto the door after that shoeprint had been made. Dunn's shoeprints in Torres's blood were found on the second step of the staircase, and a pool of blood on the pavement at the bottom of the staircase contained Dunn's shoeprint. In addition, Torres's blood was found on the sole of Dunn's shoe. Blood was found on Dunn's shorts, and DNA testing revealed a mixture of a major and minor profile, with Dunn as the source of the major profile and no conclusion able to be drawn from the minor profile.

Finally, as to the blood spatter, there was substantial evidence of several people stomping around the crime scene. Dunn was seen standing next to Torres trying to forcibly pull him up by his arm. And both Dunn and Willie were seen standing near Torres, walking next to Torres, and traversing up and down the stairs. At least three officers were seen walking around the crime scene near Torres, and photos also revealed medics walking through the blood. Moreover, while the crime scene was being processed, it began to rain.

II.

On post-conviction appeal in the state court, Dunn argued that he was denied effective assistance of counsel because his lawyer failed to consult a forensic pathologist to testify as to the cause of death, which was the pivotal issue in his criminal trial. An evidentiary hearing was held in that post-conviction proceeding, at which Dunn's trial counsel and a forensic pathologist, Dr. Thomas Sozio, testified.

At that evidentiary hearing, Dr. Sozio testified that he is a board-certified forensic pathologist, who has completed approximately 4500 autopsies in his career, and who has testified as an expert pathologist on the cause or manner of death throughout the State of Indiana, usually on behalf of the state. He testified that the autopsy conducted by Dr. Chrenka was substandard and missed a great deal. As to the critical issue at trial—the manner of death—he testified, contrary to the testimony presented by the state from Dr. Wagner, that Torres's injuries were much more consistent with a fall than with being bludgeoned by a blunt object. He explained that he observed just one small laceration on Torres's scalp, whereas if he were hit multiple times, one would expect multiple lacerations. He also noted an absence of defensive injuries. He testified that based on the area of the injury to Torres's head, and the nature of the fracture as a linear fracture rather than a depressed fracture, the fracture was more indicative of a fall. He further concluded that:

To me, all the injuries are on the same side of the body in a line. All the ribs, the lungs, the—the liver, the clavicle, all of it's the same line, significant force. That—that tells me all day everyday it's a fall.

PCR Tr. at 52.

He additionally pointed to Torres's blood alcohol content of .294, and his fatty liver consistent with chronic alcoholism, which itself is associated with reduced clotting times and greater susceptibility to broken bones. He determined that the manner of death should properly be classified as "undetermined" because the cause of the fall was undetermined, stating that "whatever occurred at the top of the stairs ... [i]t could have been a push, it could have been a strike, but all the injuries that I'm observing are basically due to a fall from height." Id. at 40. He therefore testified that he did not believe that Torres was struck at the bottom of the stairs. The state has denied reliance on the fall as the cause of the death in this case, and therefore his testimony directly refuted the state's case.

Dunn's trial counsel also testified at the evidentiary hearing. He conceded at the hearing that he was focused on the bat and evidence that the bat did not have Torres's blood on the barrel, and that focus clouded his judgment. He acknowledged that he never consulted a pathologist or blood spatter expert and did not depose Dr. Chrenka or Dr. Wagner even though the rules of the State of Indiana permitted such depositions in criminal cases. He testified that there was no strategic reason for not consulting an expert forensic pathologist in the case.

The Indiana Court of Appeals affirmed Dunn's conviction and sentence, and the Indiana Supreme Court denied his petition for transfer. The district court granted Dunn's pro se habeas petition. In so doing, the district court first held that the Indiana Court of Appeals improperly applied an incorrect legal standard, and therefore its decision was not entitled to the usual deference mandated by the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA). Applying the Strickland standard de novo , the district court then concluded that Dunn had established the deficient performance of trial counsel and prejudice. III.

Our review begins, then, with the question as to whether AEDPA deference is proper when analyzing the decision of the Indiana Court of Appeals. The AEDPA revised the standards for evaluating the merits of a habeas petition, providing under 28 U.S.C. § 2254(d) that, for a claim adjudicated on the merits in state court, "a federal court cannot grant relief unless the state court (1) contradicted or unreasonably applied [the Supreme] Court's precedents, or (2) handed down a decision ‘based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.’ " Shoop v. Twyford , ––– U.S. ––––, 142 S. Ct. 2037, 2043, 213 L.Ed.2d 318 (2022). The AEDPA restricts the consideration of new evidence by the federal habeas court, and expressly limits review of factual determinations to evidence presented in the state court proceeding. Id. ; Shinn , 142 S. Ct. at 1739. Here, the district court limited its consideration to the facts determined in the state court, including those developed in the evidentiary hearing in the state post-conviction court, and we base our decision on that record as well.

The district court in this case held that the state court decision was "contrary to" federal law, and therefore was not entitled to the usual AEDPA deference and should be reviewed de novo. As the Supreme Court explained in Williams v. Taylor , 529 U.S. 362, 405–06, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) :

A state-court decision will certainly be contrary to our clearly established precedent if the state court applies a rule that contradicts the governing law set forth in our cases. Take, for example, our decision in Strickland v. Washington , 466 U.S. 668 [104 S.Ct. 2052, 80 L.Ed.2d 674] (1984). If a state court were to reject a prisoner's claim of ineffective assistance of counsel on the grounds that the prisoner had not established by a preponderance of the evidence that the result of his criminal proceeding would have been different, that decision would be "diametrically different," "opposite in character or nature," and "mutually opposed" to our clearly established precedent because we held in Strickland that the prisoner need only demonstrate a "reasonable probability that ... the result of the proceeding would have been different." Id. at 694 .... [In that scenario], a federal court will be unconstrained by § 2254(d)(1) because the state-court decision falls within that provision's "contrary to" clause.

The error of law identified by the Supreme Court in Williams is precisely the error present in the state court decision in this case, in that the Indiana appellate court repeatedly identified the standard as requiring that the result of the proceeding would have been different. As detailed below, the Indiana appellate court in this case thus applied a rule that contradicts the governing Supreme Court law, and its decision is not entitled to AEDPA deference.

A.

There are two problems with the state court's recitation of the standard in this case. First, the court repeatedly misrepresented the standard in the manner that the Court identified as unacceptable in Williams. At critical points in its opinion, the state court characterized the standard as whether the evidence establishes that consulting with another expert would have changed the outcome of the trial. The state court did not exclusively mischaracterize the standard, reciting at times the proper standard, as to whether the evidence demonstrated a "reasonable probability" that, but for the deficient conduct, the result of the proceeding would have been different. In a number of cases, we have held that where a state court properly identifies the standard, its later failure to include language such as the "reasonable probability" limitation will not be deemed to evidence the application of a wrong standard, but instead will be interpreted to constitute merely a shorthand reference to the correct standard. See, e.g., Olvera v. Gomez , 2 F.4th 659, 671 (7th Cir. 2021) ; Malone v. Walls , 538 F.3d 744, 758 (7th Cir. 2008). That approach is consistent with our deference to state courts, and our presumption that a state court, having identified the proper standard, is then applying that proper standard in its further truncated recitation of the standard. Olvera , 2 F.4th at 671. That deferential, and practical, interpretation of a state court's opinion is entirely appropriate absent evidence that the state court did, in fact, apply a wrong standard. Id.

But as the district court found, the state court opinion in fact applied an improper standard. Its language employing the wrong standard cannot be read as a truncated version of the proper standard given the court's explanation of its role on appeal and the analysis used by the court. First, the Indiana appellate court made clear its understanding of its role at the outset of the opinion. The court stated that the post-conviction court's order denying Dunn's post-conviction petition held, in relevant part:

[Dunn] has failed in his burden to develop evidence that establishes that consulting with another expert would have changed the outcome of the trial. The witness presented by [Dunn] simply provided another expert opinion as to the cause of death of the victim which lacks some credibility when considered in connection with the totality of the evidence presented at trial.

[emphasis added] Ind. App. Mem. Decision at 12–13. The Indiana appellate court then declared that "[t]o succeed in his appeal, Dunn must convince us that ‘the evidence as a whole leads unerringly and unmistakably to a decision opposite that reached by the post-conviction court.... Put another way, Dunn ‘must convince this Court that there is no way within the law that the court below could have reached the decision it did.’ " Id. at 13. In affirming the decision of the post-conviction court, the Indiana appellate court never took issue with the post-conviction court's recitation of the burden as providing "evidence [that] establishes that consulting with another expert would have changed the outcome." That is precisely the standard which the Williams Court identified as " ‘diametrically different,’ ‘opposite in character or nature,’ and ‘mutually opposed’ to our clearly established precedent because we held in Strickland that the prisoner need only demonstrate a ‘reasonable probability that ... the result of the proceeding would have been different,’ " and which therefore falls within the "contrary to" clause of the AEDPA. Williams , 529 U.S. at 406, 120 S.Ct. 1495. Although the Indiana appellate court quoted the Strickland standard in subsequent boilerplate recitation of the requirements of the Sixth Amendment and Strickland , in its analysis portion in which it actually applied the standard, it again returned to that improper standard used by the post-conviction court. In fact, its legal conclusion finishes where it started as to the wrong standard, holding "[b]ased on this record, we cannot say that the evidence as a whole leads unerringly and unmistakably to the conclusion that the evidence produced from consulting with a forensic pathologist would have changed the outcome of Dunn's trial." Ind. App. Mem. Decision at 19. The Indiana appellate court therefore not only anchored its review in the improper standard used by the post-conviction court, but it also articulated that same improper standard itself in its conclusion and reflected that burden in its analysis. We cannot therefore dismiss the improper standard as a mere shorthand reference to the proper standard.

B.

Moreover, even if we were to determine that the failure to include the "reasonable probability" language at those critical junctures was merely a shorthand reference to the proper standard, the court's opinion is "contrary to law" for an additional reason based on the burden it imposed to demonstrate whether the evidence "would have changed the outcome." That phrase has consistently been defined as requiring only a showing of a reasonable probability that at least one juror would possess a reasonable doubt. See, e.g., Buck v. Davis , ––– U.S. ––––, 137 S. Ct. 759, 776, 197 L.Ed.2d 1 (2017) ( Strickland requires a showing of "a reasonable probability that ... at least one juror would have harbored a reasonable doubt."); Strickland , 466 U.S. at 695, 104 S.Ct. 2052 ("When a defendant challenges a conviction, the question is whether there is a reasonable probability that, absent the errors, the factfinder would have had a reasonable doubt respecting guilt."); Thomas v. Clements , 789 F.3d 760, 773 (7th Cir. 2015) (The question under Strickland is whether the omitted testimony would have been "sufficient to raise a reasonable doubt and therefore show prejudice."). Thus, a defendant need not prove his actual innocence, but need only establish reasonable doubt in the mind of at least one juror in order to change the outcome of the trial. The Indiana appellate court in its opinion never set forth that proper understanding of the requirement, however, and imposed a much more severe burden. As to that issue, the court held that:

[a]lthough Dr. Sozio's testimony may have been helpful to the defense's theory of the case, when viewed in conjunction with the totality of the evidence at trial, his testimony is not so compelling that there is a reasonable probability that had it been offered the jury would have concluded that Torres's injuries were solely the result of a fall.

[emphasis added] Ind. App. Mem. Decision at 17.

Therefore, in order to meet the requirement of showing a reasonable probability that "the outcome would have been different," the court required Dunn to establish that a jury would have concluded that the fall was the sole cause of the injuries—and thus that Dunn was innocent. In order to demonstrate a reasonable probability that evidence would have changed the outcome, however, a defendant need not convince jurors of his version of events; the defendant must merely create a reasonable doubt as to whether the government has established its version. That is not a mere semantic difference. It is fundamental that the defendant does not bear the burden of proof to demonstrate his innocence. Here, Dunn needed only to create doubt as to how the injuries were sustained. If the evidence was sufficient to raise a reasonable doubt in the mind of one juror as to how the injuries could have been sustained, that is sufficient to change the outcome. The state court, however, required Dunn to demonstrate a reasonable probability that "the jury" presented with that evidence would have concluded the injuries were "solely the result of a fall." That burden is more analogous to the stricter one applied in cases requiring a showing of actual innocence, in which, in contrast to the Strickland prejudice inquiry, the petitioner "must show that it is more likely than not that no reasonable juror would have convicted him in light of the new evidence." Schlup v. Delo , 513 U.S. 298, 327, 115 S.Ct. 851, 130 L.Ed.2d 808 (1995). That shifting of the burden is incompatible with the proper Strickland standard of establishing prejudice, and with the countless cases recognizing that the defendant need only demonstrate a reasonable probability that at least one juror will have reasonable doubt as to the state's case.

To be clear, the problem here is not that the language referred to "the jury" rather than "one juror." The dissent focuses on that, arguing that we are unreasonably demanding specific language of courts when we ourselves fail to meet that standard. That, however, misses the point. The problem here is the burden imposed by the court in the determination as to whether the outcome would be different. The court required Dunn to demonstrate a reasonable probability that the jury would conclude the injuries were solely the result of the fall. That is much more than inquiring as to whether Dunn has created a "reasonable doubt;" it requires Dunn to establish a reasonable probability that the jury would have concluded that his version of the facts is true, not that the state failed to make its case or that the jury would be unable to discern which version was true. The state court's reference to "the jury" rather than "one juror" is reflective of that misplaced burden. Dunn was essentially required to show actual innocence, a standard inapplicable to this context, but one that would focus on the impact on the jury rather than one juror (and therefore that improper language is merely reflective of the improper burden).

The analysis that followed that recitation of the improper standard by the state court reflected that the court in fact imposed that elevated burden, as illustrated by each of the state court's points that followed:

First, Dr. Sozio's testimony is not as definitive as Dunn suggests. Notably, Dr. Sozio did not conclusively identify the manner of Torres's death as an accident.... In addition, while Dr. Sozio's testimony would have provided some support for Sim's and Willie's testimony, his testimony was not definitive and their testimony suffered credibility issues.... Finally, there were facts that supported the State's common-sense argument that Torres's injuries could not have been caused merely from the fall.... Based on this record, we cannot say that the evidence as a whole leads

unerringly and unmistakably to the conclusion that the evidence produced from consulting with a forensic pathologist would have changed the outcome of Dunn's trial.

Ind. App. Mem. Decision at 17–19.

The language used in applying the standard reflects the improper burden imposed by the state court. The court did not determine that the omitted expert testimony would not have been relevant, or was cumulative, or even that the state's evidence was overwhelming regardless of that potential expert testimony. Those would have been proper considerations in assessing Strickland prejudice. The court instead discussed whether Dunn's evidence was "definitive" or "conclusive" and rejected it because although it provided "some support" it is not "conclusive" and there were facts "supporting the State's common-sense argument." That would be the proper analysis if Dunn had the burden of establishing that a jury would have found the injuries were caused solely by a fall. That was not his burden in establishing prejudice under Strickland. Reasonable doubt can be established with less than "definitive" or "conclusive" evidence; it requires only evidence that could create reasonable doubt—here doubt as to the actual cause of the injury.

The dissent agrees that it was "not Dunn's burden to prove his actual innocence through an alternative theory for Torres's death" such that the injuries were solely the result of a fall. Dissent at 714–15. But it dismisses that problem by arguing that "there were really only two possible causes of Torres's death," that Torres died from the fall or Dunn killed him, and that "[i]f there was no reasonable probability that Dr. Sozio's testimony could convince a juror that Torres's injuries ‘were solely the result of a fall,’ then it had no likelihood of affecting the trial's outcome." Id. That argument fails to recognize the obvious third option—which frankly given the evidence in this case is the most likely one—that the jury could conclude that the evidence as to the cause of death is inconclusive and is not convincing enough to establish beyond a reasonable doubt that Torres's death was the result of a beating as opposed to the fall. Equipoise is sufficient. In fact, a reasonable probability of doubt in the mind of just one juror is sufficient. The state court's opinion reflects an improper understanding of the prejudice inquiry under Strickland , and the imposition of a burden more onerous than that required by Strickland. Because the state court imposed a burden contrary to law, its decision is not entitled to deference under the AEDPA, and we consider the Strickland challenge de novo.

For completeness, we note that even absent this holding, AEDPA deference would have applied only as to the issue of prejudice under Strickland. Because the state court limited its holding to the prejudice prong and did not address deficient performance, AEDPA deference would have been inapplicable as to that issue even absent the above analysis. See e.g., Dunn v. Jess , 981 F.3d 582, 591 (7th Cir. 2020) ("AEDPA deference only applies to issues that the last reasoned state court decision reached on the merits.").

IV.

Turning to the claim on its merits, we agree with the district court that Dunn has sufficiently demonstrated ineffective assistance of counsel. A petitioner claiming ineffective assistance of trial counsel under Strickland must demonstrate: (1) that trial counsel provided constitutionally-deficient performance, "meaning counsel made errors so serious he ‘was not functioning as the "counsel" guaranteed the defendant by the Sixth Amendment; ’ " and (2) that the deficient performance prejudiced his defense, "meaning there is a ‘reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different.’ " Winfield v. Dorethy , 956 F.3d 442, 451–52 (7th Cir. 2020), quoting Strickland , 466 U.S. at 687, 694, 104 S.Ct. 2052.

In order to succeed on a claim of deficient performance, Dunn must demonstrate that his counsel's representation fell below an objective standard of reasonableness. Strickland , 466 U.S. at 688, 104 S.Ct. 2052. In making that assessment, "the court should keep in mind that counsel's function, as elaborated in prevailing professional norms, is to make the adversarial testing process work in the particular case." Strickland , 466 U.S. at 690, 104 S.Ct. 2052. In determining whether a counsel's challenged conduct is the result of reasonable professional judgment, our scrutiny is highly deferential, cognizant of the distorting effects of hindsight. Id. at 689, 104 S.Ct. 2052. We therefore employ a presumption that the challenged conduct might be considered sound trial strategy. Id. A decision not to investigate and present expert testimony as a matter of trial tactics can fall within the range of reasonable performance. Rogers v. Israel , 746 F.2d 1288, 1294 (7th Cir. 1984). "But for this deference to apply, the decision must be—in fact—strategic," and " ‘consequences of inattention rather than reasoned strategic decisions are not entitled to the presumption of reasonableness.’ " Dunn , 981 F.3d at 591, quoting Mosley v. Atchison , 689 F.3d 838, 848 (7th Cir. 2012). As the Court recognized in Strickland , "counsel has a duty to make reasonable investigations or to make a reasonable decision that makes particular investigations unnecessary," and accordingly, "[i]n any ineffectiveness case, a particular decision not to investigate must be directly assessed for reasonableness in all the circumstances, applying a heavy measure of deference to counsel's judgments." Strickland , 466 U.S. at 691, 104 S.Ct. 2052.

As in Dunn (a case only coincidentally bearing the same name as the petitioner), the failure to investigate here was attributable to inattention rather than a reasoned strategic judgment. In fact, Dunn's defense counsel in this case acknowledged that his failure to investigate by consulting a forensic pathologist was an oversight and not a strategic decision. According to his defense counsel, the failure to pursue that avenue was a product of his narrow focus on the bat as the putative weapon in the case. The critical issue in this case was whether Torres's death was caused by the fall or whether it was caused by being bludgeoned. The evidence regarding the bat—which was the focus of the defense here—is relevant to the criminal case, because the elimination of the bat as a weapon used to beat Torres made the prosecutor's case more difficult given that no other weapon had been found. But the pivotal issue in the criminal case was the cause of death, not what weapon was used—specifically, whether Torres died from being bludgeoned after the fall as the state contends or whether his injuries could be attributed to the fall itself as the defense argues. The entire case turns on proving there was no post-fall beating. Evidence that the bat was not used to beat Torres would further the defense claim if the jury had reason to believe that the fall alone could have caused the injuries, because the absence of Torres's blood on the barrel of the bat would support that argument. But absent evidence that the injuries were consistent with a fall, and with the state introducing evidence that a fall could not have caused the injuries and that the injuries were attributable to a beating, the evidence regarding the bat is less significant and certainly insufficient to refute the core issue as to the cause of death. Dunn's defense counsel did not consult with any forensic pathologist to establish the manner of death, nor did he even investigate that critical issue such as by deposing the state's expert witnesses.

In both Dunn and Rogers , we faced analogous facts in which the pivotal issue was the timing of the fatal injuries, for which the testimony of a forensic pathologist would be critical. Dunn , 981 F.3d at 595 ; Rogers , 746 F.2d at 1293. In those cases, we held that the failure to pursue the expert testimony constituted deficient performance. For instance, in Dunn , the defendant had slapped the victim causing him to fall, but the victim was able to walk away following the incident and was found dead only later with evidence of a skull fracture and significant bleeding. The critical issue for the defense was whether the injuries apparent in the autopsy would have caused immediate death, therefore negating the incident with the defendant as the cause of death. Dunn , 981 F.3d at 595. His counsel failed to explore evidence from experts that would have countered the opinion of the medical examiner. That expert testimony from a forensic pathologist would have established that the victim was never upright—and thus could not have walked away—after sustaining the injuries that proved fatal, and therefore would have established that the incident with the defendant was not the cause of the death. Id. at 591. In Dunn , the defense counsel's performance was deemed deficient because the failure to investigate resulted from inattention rather than strategic judgment, and the defense approach to focus on the nature of the injuries lent some support to the argument that the defendant did not cause the death, but was not nearly as strong as evidence that could have been supplied by an expert as to the timing of the death. See also Rogers , 746 F.2d at 1293–94 (holding that where the critical issue at trial was whether the injury to the victim would have caused his immediate incapacitation, there was a reasonable probability that even expert testimony not phrased in unconditional terms, if presented by defense counsel, could have caused the jury to have reasonable doubt respecting guilt).

The same reasoning that controlled in Dunn and Rogers applies here. Dunn's defense counsel conceded that the decision not to consult with a forensic pathologist was not a strategic decision, and that he was focused on whether the bat was used to beat Torres and that impaired his judgment. As in Dunn and Rogers , defense counsel here failed to focus on the critical argument, in that counsel failed to investigate and present evidence as to the cause of death. Only forensic evidence would have tended to rule out bludgeoning (whether by the bat or some other object) as a cause of death. And that, more critical, argument was not made because Dunn's defense counsel did not investigate to obtain evidence related to the cause of death, including consulting with a forensic pathologist, even after becoming aware that the state was producing its own expert on the issue. See Dunn , 981 F.3d at 594 (" ‘Even if defense counsel could have initially believed expert testimony unnecessary,’ finding out the medical examiner was going to testify that [the victim] did not die immediately should ‘have alerted any reasonable attorney to the need to rebut with a defense expert.’ "), quoting Woolley v. Rednour , 702 F.3d 411, 423 (7th Cir. 2012). Because defense counsel, due to inattention rather than any strategic purpose, failed to investigate the defense as to the dispositive issue at trial regarding the cause of death, Dunn has demonstrated deficient performance under Strickland. That brings us to the second prong of Strickland , that of demonstrating prejudice. In order to demonstrate prejudice, Dunn must show that "there is a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different." Strickland , 466 U.S. at 693–94, 104 S.Ct. 2052. As we have stated, a reasonable probability is probability sufficient to undermine confidence in the outcome—that is, a reasonable probability that, absent the errors, at least one factfinder would have had a reasonable doubt respecting guilt. Buck , 137 S. Ct. at 776 ; Strickland , 466 U.S. at 695, 104 S.Ct. 2052 ; Thomas , 789 F.3d at 773.

Here, no one testified that they observed Dunn hit Torres after the fall. The only evidence of a beating post-fall is the expert testimony. The state's argument from the outset was that the testimony of the eyewitnesses was not believable and that the jury should believe the scientific evidence. For instance, pursuant to Indiana rules the parties were allowed to make opening arguments of three minutes or less to the entire jury venire before jury selection. The state argued:

You will hear medical evidence that suggests that Angel Torres was severely beaten and that this was not the result of a simple fall down the stairs as the defendant reported to the police. The question for you will become: What do you believe? Do you believe the medical, scientific, and forensic evidence; or do you believe people who perhaps have reasons not to be forthcoming with you as they should be?

Trial Tr. 27. That strategy remained unchanged throughout the trial, and at closing argument the state denied reliance on the bat as the instrument used, but again argued:

If Jamar, Willie Sims, is really and truly telling you the truth, then every bit of science, every bit of gravity, every bit of physics you have to disbelieve.... People lie. It happens. And, you know, we can use terms like you don't find swans in a sewer. This are where they're living, the lifestyle they're living, we can't pick our witnesses.... But you have to compare what you learned from Letha and Jamar with what you learned from Sergeant Dean Marks and Dr. Wagner, and then you have to determine what makes more sense.

Trial Tr. 1013, 1018.

If defense counsel had presented the testimony of Dr. Sozio, then the jury would have been presented with conflicting expert testimony regarding whether the fall alone caused the injuries. The blood evidence already effectively ruled out any argument that the bat was used to beat Torres, and the state conceded as much by arguing at closing that a different, unidentified weapon was probably used. But the timeline makes that a difficult argument, because the police arrived within eight minutes of the 911 call, and that call appears to have been made during the initial argument between Sims and Dunn, which was followed by the argument with Torres, the retreat back into the apartments, the return by Dunn to Torres's apartment, and the argument on the balcony in which Torres was hitting Dunn with the bat and then fell down the stairs and over the railing to the ground. Under the state's theory, in that time Dunn would have had to also hide the weapon, and no such weapon was ever found. And the two eyewitnesses testified consistently at the scene and at trial that Torres was not beaten after the fall.

When that evidence is then combined with the expert testimony confirming that the injuries were consistent with a fall, there is certainly a reasonable probability that the jury would have some doubt as to whether Torres was in fact beaten after the fall by Dunn. The state argues that the blood spatter evidence was uncontradicted and that Dr. Sozio's opinion, which refuted those blood spatter conclusions, would not be admissible. But even setting aside Dr. Sozio's contrary opinion regarding that evidence, the state's blood spatter argument would be heard in the context of the blood evidence tied to Dunn. Blood from Torres was found on Dunn's shoe, as would be expected given the testimony that he ran down to Torres after the fall and pulled on him imploring him to get up. But as the district court noted, the evidence did not reveal any blood on Dunn's shorts that could be identified as Torres's blood. If, as the state argues, the weapon was so saturated by blood that it caused the blood spatter when swung, the absence of evidence of blood spatter tied to Torres on Dunn's clothes diminishes the argument that he wielded such a weapon. The testimony by Dr. Sozio was critical in this case to creating reasonable doubt, because it countered the scientific evidence by the state and gave the jury reason to doubt that Torres was beaten and to believe the testimony of the defense witnesses that Torres's injuries resulted from a fall. With that evidence, there is a reasonable probability that the jury would have reasonable doubt as to Dunn's guilt, and therefore Dunn has demonstrated prejudice under Strickland . See Strickland , 466 U.S. at 695–96, 104 S.Ct. 2052 ("a verdict or conclusion only weakly supported by the record is more likely to have been affected by errors than one with overwhelming record support"); Thomas v. Clements , 789 F.3d 760, 772 (7th Cir. 2015) (holding that the failure to present affirmative expert testimony that the autopsy results were not inconsistent with the defendant's statement constituted prejudice).

V.

Accordingly, the decision of the district court granting the petition for a writ of habeas corpus is AFFIRMED.

Kirsch, Circuit Judge, dissenting.

A federal court may disturb a final state-court conviction in only a few, narrow circumstances. In the majority's view, one such circumstance applies because the Indiana Court of Appeals' decision was "contrary to" the Supreme Court's decision in Strickland v. Washington , 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). Because the Indiana Court of Appeals reasonably applied Strickland 's prejudice prong, I disagree with that conclusion and respectfully dissent.

"[T]he writ of habeas corpus is an extraordinary remedy that guards only against extreme malfunctions in the state criminal justice systems." Shinn v. Ramirez , ––– U.S. ––––, 142 S. Ct. 1718, 1731, 212 L.Ed.2d 713 (2022) (citation omitted). The Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA) "imposes several limits on habeas relief" to ensure that the writ retains this circumscribed role. Id. The relevant AEDPA limit here reads:

An application for a writ of habeas corpus on behalf of a person in custody pursuant to the judgment of a State court shall not be granted with respect to any claim that was adjudicated on the merits in State court proceedings unless the adjudication of the claim ... resulted in a decision that 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 state court's decision is ‘contrary to’ clearly established federal law where it is ‘substantially different from the relevant precedent’ of the Supreme Court." Mosley v. Atchison , 689 F.3d 838, 844 (7th Cir. 2012) (quoting Williams v. Taylor , 529 U.S. 362, 405, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) ). Put another way, a state-court decision contradicts Supreme Court precedent "if the state court arrives at a conclusion opposite to" that of the Supreme Court on a question of law or "if the state court confronts facts that are materially indistinguishable from a relevant Supreme Court precedent and arrives at a result opposite to [the Court's]." Williams , 529 U.S. at 405, 120 S.Ct. 1495. "AEDPA's requirements reflect a presumption that state courts know and follow the law." Woods v. Donald , 575 U.S. 312, 316, 135 S.Ct. 1372, 191 L.Ed.2d 464 (2015) (per curiam) (citation omitted). And "[a] state court's determination that a claim lacks merit precludes federal habeas relief so long as fairminded jurists could disagree on the correctness of the state court's decision." Harrington v. Richter , 562 U.S. 86, 101, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011) (citation omitted).

The majority thinks this demanding standard has been met because the Indiana Court of Appeals "mischaracterize[d] the standard" set out in Strickland v. Washington , which protects a criminal defendant's Sixth Amendment right to competent representation. Yet the majority's focus is misplaced. Because a "state court need not cite or even be aware of [Supreme Court] cases under § 2254(d)," the proper inquiry focuses not on how the state court characterizes a relevant Supreme Court precedent but on how it actually applies the governing law. Harrington , 562 U.S. at 98, 131 S.Ct. 770 ; Olvera v. Gomez , 2 F.4th 659, 671 (7th Cir. 2021) ("Absent circumstances that would raise a grave concern that the state court actually applied a contrary standard," we will not "override the benefit of the doubt that § 2254 provides to the state court's decision.") (emphasis added).

But even on the majority's own terms, the Indiana Court of Appeals correctly characterized Strickland. To establish a violation under Strickland , a prisoner must show "a reasonable probability" that, "but for counsel's unprofessional errors, the result of the proceeding would have been different." 466 U.S. at 694, 104 S.Ct. 2052. Strickland has been the law for nearly 40 years, and it's evident that the Indiana appellate judges deciding Dunn's petition in October 2017 (who had a combined seven decades of judicial experience, Judge Crone (since 2004), Judge Vaidik (since 1992), and Judge Mathias (since 1989)) correctly understood and recited Strickland 's prejudice prong. See Sussman v. Jenkins , 636 F.3d 329, 359–60 (7th Cir. 2011) (looking to whether it was "clear from the [state] court's analysis that it did not believe that the [proffered evidence] had a reasonable probability of altering the jury's verdict"). Indeed, the court stated the standard correctly at least four times in its decision denying Dunn post-conviction relief. See Indiana App. Mem. Decision at 14 ("To establish prejudice, the petitioner must show that there is a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different."); id. at 15 (framing issue as "whether, but for counsel's unprofessional conduct, there was a reasonable probability that a different verdict would have been reached"); id. at 16 ("[S]uccess on the prejudice prong of an ineffectiveness claim requires a showing of a reasonable probability of affecting the result.") (citation omitted); id. (describing issue as "whether the evidence produced from consultation with a forensic pathologist would have had a reasonable probability of affecting the outcome of Dunn's trial"). Dismissing these four accurate statements of the standard as "boilerplate," the majority concludes that the Indiana Court of Appeals improperly characterized Strickland 's prejudice prong. It starts by inferring that the Indiana Court of Appeals adopted the lower court's reasoning by quoting the court's order in the "Case Summary" section of its opinion (the Court of Appeals' "Discussion and Decision" came later). But we do not ordinarily treat a court's recitation of a case's background and procedural history as illuminating. In the majority's view, the Indiana Court of Appeals needed to affirmatively "t[ake] issue" with the quoted language of the order under its review. Yet we often refine the legal analysis of orders we affirm without correcting all potentially problematic reasoning contained within those orders. I hesitate to think how many of our own opinions would fail to pass muster if the Supreme Court required us to expressly disavow all improper lower court reasoning before affirming a judgment. In any event, the majority's inference contradicts our duties to " ‘presum[e] that state courts know and follow the law’ and [to] give their articulation of that standard ‘the benefit of the doubt.’ " Olvera , 2 F.4th at 670 (quoting Woodford v. Visciotti , 537 U.S. 19, 24, 123 S.Ct. 357, 154 L.Ed.2d 279 (2002) (per curiam)).

The majority next implies that the Indiana Court of Appeals erred by reciting the applicable standard of review for post-conviction appeals in Indiana in its Strickland analysis. As described by the Indiana Supreme Court, post-conviction petitions on appeal "must convince the court that the evidence as a whole leads unerringly and unmistakably to a decision opposite that reached by the post-conviction court." Wesley v. State , 788 N.E.2d 1247, 1250 (Ind. 2003). Contrary to the majority's implicit suggestion, then, this "unerring and unmistaken" appellate standard had nothing to do with the Indiana Court of Appeals' understanding of Strickland. Instead, it involved Indiana's procedural division between appellate and trial court functions, a division we are bound to respect on habeas review. See Trevino v. Thaler , 569 U.S. 413, 421, 133 S.Ct. 1911, 185 L.Ed.2d 1044 (2013) (noting "the importance of federal habeas corpus principles designed to prevent federal courts from interfering with a State's application of its own firmly established, consistently followed, constitutionally proper procedural rules").

After the smoke from these irrelevant explanations dissipates, the majority's issue with the Indiana Court of Appeals' framing of the Strickland standard largely comes down to one sentence in the Court of Appeals' concluding paragraph:

Based on this record, we cannot say that the evidence as a whole leads unerringly and unmistakably to the conclusion that the evidence produced from consulting with a forensic pathologist would have changed the outcome of Dunn's trial.

Of course, I agree with the majority that, standing alone, this sentence does not accurately state the Strickland standard. But when viewed as a whole and in context, this single misstatement does not make the decision "contrary to" Strickland. The Indiana Court of Appeals' opening paragraph is nearly identical to its conclusion, with the six-word difference underlined as follows:

Based on the record before us, we cannot say that the evidence as a whole leads unerringly and unmistakably to the conclusion that there is a reasonable probability that the evidence produced from consulting with a forensic pathologist would have changed the outcome of Dunn's trial.

If the Indiana Court of Appeals had repeated these six words in its conclusion, we would not be here. And as the majority recognizes, we have repeatedly held that a state court's use of incomplete or incorrect shorthand of the Strickland standard in a conclusion does not make its decision "contrary to" Strickland , so long as the rest of the analysis applies the correct standard. See, e.g., Olvera , 2 F.4th at 670 (holding that state appellate court's statement that "we cannot find this new evidence would have resulted in an acquittal" was not contrary to Strickland ); Gage v. Richardson , 978 F.3d 522, 528 (7th Cir. 2020) ("When a state court misstates Strickland 's prejudice prong, AEDPA deference may still apply if its analysis focused on whether the proffered testimony could have affected the outcome," "even when the incorrect recitation seemingly places an additional burden on the petitioner.") (citation omitted); Fayemi v. Ruskin , 966 F.3d 591, 594 (7th Cir. 2020) (state court's use of the phrase "would likely have been different" twice was not contrary to Strickland ); Sussman , 636 F.3d at 359–60 (holding that a state court's omission of the "reasonable probability" language when defining prejudice was not contrary to Strickland ); Woods v. Schwartz , 589 F.3d 368, 378 n.3 (7th Cir. 2009) ("We have noted numerous times that there is no error when a court has correctly noted the Strickland standard and then used an incorrect shorthand version when stating its conclusion."); Malone v. Walls , 538 F.3d 744, 758 (7th Cir. 2008) (state court's statement of prejudice standard as "the outcome of the trial would have been different" was not contrary to Strickland ); Stanley v. Bartley , 465 F.3d 810, 813 (7th Cir. 2006) (omission of "reasonable probability" language was not contrary to Strickland ). As in these on-point precedents, the Indiana Court of Appeals' misstated sentence is merely that: a misstatement.

Even if I'm correct on this point, the majority still thinks the Indiana Court of Appeals mischaracterized Strickland 's prejudice prong. In the majority's view, the decision needed to "set forth th[e] proper understanding" that prejudice requires only that there be a reasonable probability that at least one juror—and not "the jury," as in the entire jury—would possess a reasonable doubt had counsel performed effectively. Although this articulation is no doubt correct, appellate courts don't need to include that language in every Strickland case to avoid contradicting the Supreme Court's decision (after all, Strickland itself did not employ that language). Indeed, the Supreme Court is often not that explicit, see, e.g., Andrus v. Texas , ––– U.S. ––––, 140 S. Ct. 1875, 1885–86, 207 L.Ed.2d 335 (2020) (per curiam) ("Here, prejudice exists if there is a reasonable probability that, but for his counsel's ineffectiveness, the jury would have made a different judgment about whether Andrus deserved the death penalty as opposed to a lesser sentence.") (emphasis added), and neither are we, see, e.g., Shannon v. United States , 39 F.4th 868, 882 (7th Cir. 2022) (Hamilton, J.) ("Considering all the other evidence against Shannon, however, we find it highly unlikely that removing A.W.'s testimony from the equation—assuming that had been a reasonable probability—would have given the jury reasonable doubt about Shannon's guilt.") (emphasis added). We shouldn't hold the Indiana Court of Appeals to a higher standard on habeas review than we hold ourselves.

But even if the majority is right that the Indiana Court of Appeals could have described Strickland 's prejudice prong more sharply, we would still need to find that it actually "applied a different standard" to avoid AEDPA deference. Stanley , 465 F.3d at 813. Given the general nature of the Strickland standard, "the range of reasonable applications is substantial," Harrington , 562 U.S. at 105, 131 S.Ct. 770, and state courts have considerable "leeway ... in reaching outcomes in case-by-case determinations," Knowles v. Mirzayance , 556 U.S. 111, 123, 129 S.Ct. 1411, 173 L.Ed.2d 251 (2009) (citation omitted). And here the Indiana Court of Appeals reasonably applied Strickland by carefully assessing Dr. Sozio's testimony against the trial evidence before explaining that Dunn had not shown a reasonable probability of a different outcome based on the new testimony. First, the court did not see Dr. Sozio's testimony as persuasive because he: (1) did not "identify the manner of Torres's death as an accident" since he testified that the manner of death was "best" viewed as "undetermined"; (2) agreed that Torres (as a chronic alcoholic) could have functioned quite well at a high blood alcohol level, which undercut his suggestion that Torres's alcohol influence caused him to fall; (3) only opined that it was "highly unlikely" that Torres's rib injuries—and not all of his injuries (including those to the head)—were caused by a blunt-object strike; (4) admitted he was not a blood spatter expert; and (5) admitted that Torres's scalp laceration was on an area of the head more indicative of an assault. The Indiana Court of Appeals also explained the strength of the State's trial evidence, including: (1) the State's forensic pathologist's detailed testimony and conclusive opinion that the manner of death was homicide and that the injuries could not have been caused by a fall; (2) the State's blood spatter expert's "compelling" testimony that the blood spatter was not from a fall and that a swung or thrown object likely caused the cast-off blood spatter; and (3) the relatively low height of the balcony from which Torres allegedly fell weighed against the severity of his injuries. Last, the court considered the weaknesses in Dunn's defense, finding that his two witnesses, Sims and Willie, both suffered from credibility issues because: (1) part of their testimony differed from other witnesses near the time of Torres's injuries; (2) Sims' testimony about doing cocaine with Torres did not match Torres's toxicology report; and (3) Sims admitted that she had lied to a detective about whether Dunn had kicked Torres at the bottom of the stairs.

Despite these detailed factual explanations, the majority thinks the Indiana Court of Appeals improperly applied Strickland in the following sentence:

Although Dr. Sozio's testimony may have been helpful to the defense's theory of the case, when viewed in conjunction with the totality of the evidence at trial, his testimony is not so compelling that there is a reasonable probability that had it been offered the jury would have concluded that Torres's injuries were solely the result of a fall.

The majority points out that it was not Dunn's burden to prove his actual innocence through an alternative theory for Torres's death (that his injuries were solely caused by a fall down the stairs). That's true as far as it goes, but Dunn still had the burden to show a reasonable probability of a different trial outcome absent his counsel's allegedly ineffective performance. See Strickland , 466 U.S. at 694, 104 S.Ct. 2052. Dunn sought to use Dr. Sozio's testimony to undermine the prosecution's theory that Dunn savagely beat Torres with an unidentified object. Given the record here, there were really only two possible causes of Torres's death: either he died from a fall down the stairs or Dunn killed him. If there was no reasonable probability that Dr. Sozio's testimony could convince a juror that Torres's injuries "were solely the result of a fall," then it had no likelihood of affecting the trial's outcome. And the Indiana Court of Appeals offered multiple reasons for why Dr. Sozio's testimony could not have persuaded a reasonable juror of that possibility. Consider the blood spatter expert's testimony. Even if Dr. Sozio had testified at trial, the jury still would have heard uncontradicted expert testimony about multiple areas of blood spatter found on the building's walls, on Torres's shoulder, on the pavement, on the stair steps, on the stair railings, and on a nearby car's doors. In particular, the jury would have still heard that the "impact" blood spatter found at the wall near the bottom of the stairs took a "good amount of energy ... applied to the blood source" to create and did not result from a fall or someone walking or stomping through the pooled blood on the pavement. The jury also would have still heard that the "cast-off blood staining" found on the concrete between the stairs and the car was likely caused by a swung or thrown bloody instrument. And the jury would have still heard the blood spatter expert's testimony that an "event" occurred at the bottom of the stairs that led to the spatter. This uncontradicted evidence matches the jury's finding that Dunn killed Torres with a blunt object and critically undermines Dunn's theory that a fall caused Torres's death. Indeed, even Dunn's trial counsel admitted that a reasonable person would be "impressed" by this blood spatter testimony. Yet Dr. Sozio's testimony failed to counter this evidence. Indeed, Dr. Sozio admitted that he was not a blood spatter or bloodstain expert, and Dunn offered nothing else in his post-conviction evidentiary hearing to rebut the State's blood spatter expert. So we must presume that no such evidence exists. Cf. Shinn , 142 S. Ct. at 1739. The Indiana Court of Appeals had overwhelming support for its conclusion that Dr. Sozio's testimony had no reasonable probability of changing the outcome of Dunn's trial. See Harrington , 562 U.S. at 112, 131 S.Ct. 770 ("The likelihood of a different result must be substantial, not just conceivable."). The Indiana Court of Appeals thus did not stray from Strickland 's requirements. As a result, we lack power to disturb its holding.

Although this may have been a close case for the Indiana Court of Appeals, deciding whether the court came out on the "right" side is not our call to make so long as its explanations were reasonable. See Schriro v. Landrigan , 550 U.S. 465, 473, 127 S.Ct. 1933, 167 L.Ed.2d 836 (2007) ("The question under AEDPA is not whether a federal court believes the state court's determination was incorrect but whether that determination was unreasonable—a substantially higher threshold."). Congress has set a high hurdle in requiring us to find a state court's decision "contrary to" a clearly established Supreme Court precedent before we may grant habeas relief. And we can't clear that hurdle here. In holding that the decision reached by the Indiana Court of Appeals was "contrary to" Strickland , the majority fixates on one shorthand statement and misconstrues several others all while ignoring the court's detailed application of Strickland to the facts. Only by doing so can the majority review the Strickland issue de novo to reach a different result from the Indiana Court of Appeals. Yet this path flouts the "unwavering respect" we must afford to state court adjudications of guilt. See Shinn , 142 S. Ct. at 1739. And it shirks our duty to give effect to AEDPA's statutory commands. See id. at 1736 ("Where Congress has erected a constitutionally valid barrier to habeas relief, a court cannot decline to give it effect.") (citation omitted). Because the Indiana Court of Appeals' proper application of Strickland leaves us no room to bypass AEDPA's required deference, I respectfully dissent.