concurring in part and dissenting in part.
While I agree with the majority’s resolution of the Confrontation Clause issue, I disagree with its conclusion that the Michigan state courts’ rejection of Dendel’s ineffective assistance of counsel claim “was contrary to, or involved an unreasonable application of, clearly established Federal law” — the standard state prisoners must overcome to receive habeas relief under the Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”), 28 U.S.C. § 2254(d)(1). This standard “severely constrains” our review. Smith v. Anderson, 632 F.3d 277, 281 (6th Cir.2011). For ineffective assistance claims, the clearly established federal law is of course Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). Strickland requires a showing that counsel’s deficient performance prejudiced the defense. Id. at 687, 104 S.Ct. 2052. But on AEDPA review, “the pivotal question is *617whether the state court’s application of the Strickland standard was unreasonable. This is different from asking whether defense counsel’s performance fell below Strickland’s standard.” Harrington v. Richter, 562 U.S. 86, 101, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011). Because • the Strickland standard is a general one, the range of reasonable applications is substantial. Id. at 105, 131 S.Ct. 770. In fact, where both Strickland and § 2254(d) apply, our analysis should be “doubly deferential.” Woods v. Donald, — U.S.-, 135 S.Ct. 1372, 1376, 191 L.Ed.2d 464 (2015). In my view, the majority opinion does not adequately account for these principles and, in doing so, reaches the wrong result.
The majority finds prejudice because Dendel’s counsel failed to investigate and call an expert witness who could testify to an alternative cause of death. The Michigan Supreme Court, however, affirmed the state trial court’s conclusion that Dendel was not prejudiced by this failure because the new expert testimony presented at her post-conviction Ginther hearing did not establish a reasonable likelihood of a different result at trial. Our rule for habeas petitions is to give these state court decisions “the benefit of the doubt.” Renico v. Lett, 559 U.S. 766, 773, 130 S.Ct. 1855, 176 L.Ed.2d 678 (2010) (quoting Woodford v. Visciotti, 537 U.S. 19, 24, 123 S.Ct. 357, 154 L.Ed.2d 279 (2002) (per curiam)). The majority instead (and in spite of AEDPA) appears to have engaged in a de novo review of Dendel’s ineffective assistance claim. But it is- the state court decisions we must examine to adjudicate Dendel’s claim, not merely our own independent evaluation of the underlying evidence.
After the state trial court denied Den-del’s motion for a new trial, the Michigan Court of Appeals remanded the case for a Ginther hearing to determine whether Dendel’s trial attorney rendered ineffective assistance. See People v. Dendel, 481 Mich. 114, 748 N.W.2d 859, 863 (2008). At the Ginther hearing, the state trial judge heard new defense expert testimony from Dr. Simson. Dr. Simson contested the prosecution’s theory that Burley died of hypoglycemic shock triggered by an insulin overdose and proposed that Burley actually died of a multiple-drug (primarily morphine) overdose. Dr. Simson conceded, however, that he could not rule out the possibility that Burley died of an insulin injection. The trial court also heard the prosecution recall its expert, Dr. Pa-cris, who defended his original testimony against Dr. Simson’s critiques. Dr. Pacris also rejected a morphine overdose as the cause of death due to Burley’s history of morphine use. At the close of the hearing, the state trial judge determined that defense counsel’s performance was objectively reasonable and, furthermore, that the outcome of the trial would not have been different had the defense offered Dr. Sim-son’s testimony because “there was a lot of other evidence” besides the prosecution’s expert pointing to Dendel’s guilt. Id.
In reviewing the conflicting expert testimony presented in Dendel’s Ginther hearing, the Michigan Supreme Court reasonably concluded that Dr. Simson’s testimony would not have been substantially likely to produce a different result at trial. Id. at 865-68 & n. 17. The Michigan Supreme Court reasonably upheld the trial court’s implicit credibility finding that “Dr. Simson was not more credible than the prosecution’s experts.” Id, at 867. The court highlighted that Dr. Pa-cris strongly defended his position that Burley died of hypoglycemic shock against' Dr. Simson’s challenges. Id. The court further emphasized that Dr. Simson did not respond to Dr. Pacris’s testimony and “Dr. Simson conceded the possibility that Burley had died from insulin overdose.” Id. at 868. In addition, the court explained that Dendel herself made state*618ments to the police that Burley had injected himself with insulin, and these statements supported Dr. Pacris’s theory of death — not Dr. Simson’s. Id. Lastly, the court outlined the extensive circumstantial evidence supporting the guilty verdict, including Dendel’s inculpatory statements, her difficulty with Burley in the months leading up to his death, her opportunity and motive to inject Burley with insulin around 4 a.m. — the time Dr. Pacris testified Burley fell into a coma, and Dendel’s suspicious behavior following Burley’s death. Id. at 868-70.
The showing required to satisfy Strickland ’s prejudice prong is well-established. The question is not “whether it is possible a reasonable doubt might have been established if counsel acted differently. Instead, Strickland asks whether it is ‘reasonably likely5 the result would have different.” Harrington, 562 U.S. at 111, 131 S.Ct. 770 (internal citations omitted). “The likelihood of a different result must be substantial, not just conceivable.” Id. at 112, 131 S.Ct. 770.
It strains the notion of deference to read the Michigan state court decisions as an unreasonable application of this flexible legal standard. In fact, the Michigan Supreme Court’s thorough analysis reveals solid ground for finding a lack of prejudice. The Michigan Supreme Court assessed the newly uncovered expert testimony and reasonably concluded, based on the weight of the competing expert testimony and circumstantial evidence, that there was not a reasonable probability that Dendel would have been acquitted even with the new expert testimony. See Sears v. Upton, 561 U.S. 945, 955-56, 130 S.Ct. 3259, 177 L.Ed.2d 1025 (2010) (per curiam) (explaining that a proper analysis of prejudice under Strickland requires a “probing and fact-specific analysis” whereby the state court takes into account “the newly uncovered evidence” along with the evidence introduced during trial to assess whether there is a reasonable probability of a different result).
Despite the majority’s suggestion, the Michigan Supreme Court decision is not contrary to or an unreasonable application of Rompilla v. Beard, 545 U.S. 374, 125 S.Ct. 2456, 162 L.Ed.2d 360 (2005). The Rompilla Court found prejudice because the newly-discovered evidence in that case produced “a mitigation case that [bore] no relation to the few naked pleas for mercy actually put before the jury.” 545 U.S. at 393, 125 S.Ct. 2456. The Michigan Supreme Court, in contrast, held that Dr. Simson’s expert testimony did not substantially alter the case presented to the trial court in the form of Dr. Pacris’s expert testimony and the significant additional evidence supporting Dr. Pacris’s theory of death.
I therefore would affirm the district court’s decision denying the petition for a writ of habeas corpus.