NOT RECOMMENDED FOR PUBLICATION File Name: 25a0517n.06
No. 24-2049
UNITED STATES COURT OF APPEALS FILED FOR THE SIXTH CIRCUIT Nov 04, 2025 KELLY L. STEPHENS, Clerk
)
)
ALLEN W. KECK, )
Petitioner-Appellant, ) ON APPEAL FROM THE UNITED ) STATES DISTRICT COURT FOR THE v. ) EASTERN DISTRICT OF MICHIGAN )
DALE BONN, Warden, ) OPINION Respondent-Appellee. )
)
Before: STRANCH, BUSH, and DAVIS, Circuit Judges.
JOHN K. BUSH, Circuit Judge. In 2018, a Michigan jury convicted petitioner Allen W.
Keck of first-degree child abuse of his three-month-old daughter CK. This was Keck’s second felony conviction, the first being the 1993 second-degree murder of his three-month-old daughter TK. After a failed direct appeal in Michigan state court, Keck sought a writ of habeas corpus in federal district court, raising Sixth Amendment ineffective-assistance-of-counsel claims. The district court denied his petition. We AFFIRM.
I.
The Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA) requires that we presume the validity of the factual determinations of the last-reasoned state-court opinion in the case. 28 U.S.C. § 2254(e)(1). Here, the Michigan Court of Appeals rendered that opinion. So Keck bears the burden of rebutting the presumptive correctness of the Michigan Court of Appeals’ factual findings by clear and convincing evidence. Id. Because Keck has not meaningfully
2
contested any of those findings, we will proceed based on their veracity. See People v. Keck, No. 346077, 2022 WL 128582, at *1 (Mich. Ct. App. Jan. 13, 2022) (per curiam).
The trouble arose one day when CK’s mother, Jennifer Karaffa, went to work. While away from home she entrusted Keck to care for CK along with their other child, a twenty-month-old. When Karaffa left for her job, CK was a normal, healthy three-month-old girl. But that was not the case upon Karaffa’s return home in the early hours of the following day. When Karaffa arrived, CK was crying. The infant refused the formula her mother attempted to feed her. The baby displayed swelling and redness around her head and red bruising around her right eye. Hours later, she projectile vomited what little formula Karaffa had succeeded in feeding her. Keck denied knowing what had caused CK’s condition.
Later that day, Karaffa took her ailing girl to her pediatrician. Upon the pediatrician’s direction, Karaffa took CK to the hospital. There, the diagnosis included multiple conditions consistent with intentional physical abuse: (1) fractures on both sides of the baby’s skull, (2) bleeding below the scalp known as a subdural hematoma, (3) several recent rib fractures, (4) a past rib fracture that had subsequently healed, (5) a fractured femur, and (6) numerous retinal hemorrhages in both eyes.
Thankfully, CK survived, and an investigation ensued. Keck and Karaffa suggested CK’s injuries had resulted from their twenty-month-old sitting on his infant sister’s head a few days earlier. After being advised that explanation was impossible, the parents shifted their story, attributing CK’s injuries to the actions of Karaffa’s eleven-year-old daughter GK. Keck and Karaffa suggested GK could have caused the injuries a few weeks before by sitting on CK’s head or playfully throwing the baby up in the air and catching her. But none of this persuaded the investigators, who initiated criminal proceedings against Keck.
3
Trial lasted seven days. The prosecution largely relied on (1) several of CK’s treating physicians’ testimony and (2) evidence about Keck’s second-degree murder conviction. Keck’s defense heavily relied on expert witness Dr. Marcus DeGraw, a child-abuse pediatrician. The defense argued that CK’s injuries could have occurred as early as two weeks before they were discovered, suggesting that one of the other two children caused CK’s injuries by sitting on her or playing too roughly with her. The jury found none of this persuasive and returned a verdict that Keck was guilty of first-degree child abuse. The district court then sentenced him to 14–30 years in prison.
The Michigan Court of Appeals reviewed Keck’s conviction on the merits and affirmed.
Keck, 2022 WL 128582, at *6–11. Keck applied for leave to appeal, but the Michigan Supreme Court summarily denied review. People v. Keck, 975 N.W.2d 466 (Mich. 2022) (mem.). Keck then petitioned for a writ of habeas corpus in federal district court. Keck v. Davids, No. 4:23-CV- 12432, 2024 WL 4595118, *1 (E.D. Mich. Oct. 28, 2024). The district court denied his petition but issued a certificate of appealability. Id. at *10–11. Keck timely appealed.
II.
In habeas cases, AEDPA provides the relevant standard of review when the last-reasoned state-court decision is on the merits. 28 U.S.C. § 2254(d); Reiner v. Woods, 955 F.3d 549, 556 (6th Cir. 2020). Because the last-reasoned state-court decision in the instant case—the direct appeal of Keck’s conviction in the Michigan Court of Appeals—was on the merits, see Keck, 2022 WL 128582, at *6–11, we apply AEDPA deference.
“AEDPA erects a formidable barrier to federal habeas relief for prisoners whose claims have been adjudicated in state court.” McGowan v. Burt, 788 F.3d 510, 514 (6th Cir. 2015)
4
(quoting Burt v. Titlow, 571 U.S. 12, 19 (2013)). That formidable barrier is set out in 28 U.S.C. § 2254(d):
An application for a writ of habeas corpus . . . 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—
(1) 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; or
(2) resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.
“[A] state court decision is contrary to clearly established law only if it ‘(1) applies a rule that directly conflicts with a rule prescribed by the Supreme Court or (2) confronts a case with materially identical facts to a Supreme Court decision and decides the case differently.’” Hodge v. Plappert, 136 F.4th 648, 661 (6th Cir. 2025) (en banc) (quoting Rogers v. Mays, 69 F.4th 381, 389 (6th Cir. 2023) (en banc)). For a state court to have unreasonably applied clearly established federal law, “the state court’s determination must be ‘so obviously wrong that its error lies beyond any possibility for fair[-]minded disagreement.’” Id. (quoting Shinn v. Kayer, 592 U.S. 111, 124 (2020) (per curiam)) (cleaned up). This requires a showing far higher than even clear error—the state court decision “must be so ‘lacking in justification’” that it is “objectively unreasonable.” Id. (quoting Virginia v LeBlanc, 582 U.S. 91, 94 (2017) (per curiam)). In fact, it must be completely impossible for a fair-minded jurist to believe the state court’s decision was consistent with existing Supreme Court precedent. Id.
The foregoing is an extremely deferential standard. But that level of deference is doubled when dealing with a habeas petition premised on Sixth Amendment ineffective-assistance-of- counsel claims. Burt, 788 F.3d at 513 (citing Titlow, 517 U.S. at 15); Yancey v. Haas, 742 F. App’x 980, 982 (6th Cir. 2018).
5
We evaluate ineffective-assistance-of-counsel claims under the two-step test set out in Strickland v. Washington, 466 U.S. 668 (1984). The first step requires a petitioner to show “that [his] counsel’s representation fell below an objective standard of reasonableness.” Id. at 688. Our scrutiny of representation must be “highly deferential” to trial counsel. Id. at 689. We “indulge a strong presumption” that counsel “made all significant decisions in the exercise of reasonable professional judgment.” Id. at 689–90. The second step requires a petitioner to show prejudice— that is, the “reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Id. at 694. A reasonable probability is one “sufficient to undermine confidence in the outcome.” Id. And “[t]hat requires a ‘substantial,’ not just ‘conceivable,’ likelihood of a different result.” Cullen v. Pinholster, 563 U.S. 170, 189 (2011) (quoting Harrington v. Richter, 562 U.S. 86, 112 (2011)).
So each step of the Strickland test sets a high bar for the petitioner. The upshot is that we apply a “doubly deferential” standard in reviewing habeas petitions that invoke Strickland—our “‘highly deferential’ look at counsel’s performance” is filtered “through the ‘deferential lens of § 2254(d).’” Id. at 190 (first quoting Knowles v. Mirzayance, 556 U.S. 111, 123 (2009); then quoting Strickland, 466 U.S. at 689; and then quoting Mirzayance, 556 U.S. at 121 n.2). Under that doubly deferential standard, the question before us is whether “there is any reasonable argument that counsel satisfied Strickland’s deferential standard.” Richter, 562 U.S. at 105. Put differently, we ask if any reasonable jurist could conclude that trial counsel’s conduct was reasonable. If the answer is yes, Keck loses.
III.
Keck argues his trial counsel was ineffective in (1) relying on Dr. DeGraw’s expert testimony, (2) failing to provide retinal imaging to Dr. DeGraw ahead of trial, and (3) insufficiently
6
impeaching a prosecution witness. The district court was not persuaded by any of these arguments. Neither are we. A. Reliance on Dr. DeGraw Keck’s primary argument on appeal is that his trial counsel erred by calling Dr. DeGraw as the only expert witness for the defense. That decision is entitled to “a ‘strong presumption’ of reasonableness.” Dunn v. Reeves, 594 U.S. 731, 739 (2021) (per curiam) (quoting Richter, 562 U.S. at 104). “The selection of an expert witness is a paradigmatic example of the type of ‘strategic choice’ that, when made ‘after thorough investigation of the law and facts,’ is ‘virtually unchallengeable.’” Hinton v. Alabama, 571 U.S. 263, 275 (2014) (per curiam) (quoting Strickland, 466 U.S. at 690) (cleaned up). So it is unsurprising that the Michigan Court of Appeals rejected this argument. Keck, 2022 WL 128582, at *5, *8–11. That rejection was not objectively unreasonable.
Keck makes much ado of his trial counsel’s hindsight admission that other experts might have been helpful. But Strickland is an objective standard, 466 U.S. at 688, so Keck’s trial counsel’s subjective evaluation of his own performance is irrelevant. If there is any plausible argument that his trial counsel’s performance was sufficient, Keck’s habeas petition fails. Richter, 562 U.S. at 105. And there is clearly a plausible argument that reliance on Dr. DeGraw was reasonable. Dr. DeGraw is a highly qualified child-abuse pediatrician. He has extensive experience testifying in child-abuse cases for both the prosecution and the defense. Trial counsel could have reasonably concluded that evenhandedness would bolster Dr. DeGraw’s credibility with a jury. And both Keck and his trial counsel were comfortable and familiar with Dr. DeGraw because Dr. DeGraw testified at Keck’s parental-rights-termination hearing—in fact, Keck insisted that Dr. DeGraw be called as an expert witness at trial.
7
The reasonableness of that decision was borne out at trial because Dr. DeGraw’s testimony was helpful to Keck’s defense. Dr. DeGraw disputed the medical conclusions provided by the prosecution’s witnesses. He testified that the medical imaging did not establish CK suffered fractures to her ribs and femur. He also testified that what appeared to be a bruise on CK’s eyelid may have been a malformed blood vessel. Most importantly, Dr. DeGraw testified that CK’s head injuries could have been caused by another child sitting on her, and that those injuries were at least a week (perhaps even several weeks) old. Dr. DeGraw claimed “there really isn’t any doubt that significant blunt force trauma or crush force trauma” was responsible for CK’s most significant cranial injuries. R. 9-17, 8/10/18 Trial Tr., PageID 4445–47. He further explained that it was possible for a 100- to 150-pound object, such as a child, to fall a few feet and cause these cranial injuries. In sum, Dr. DeGraw provided expert testimony directly supporting the defense’s theory of the case and directly refuting the prosecution’s witnesses. Trial counsel’s decision to rely on this testimony was a reasonable strategic choice.
Keck makes several unpersuasive arguments in response.
He first argues that Dr. DeGraw’s testimony was actually harmful to his case because Dr. DeGraw admitted on cross that child abuse possibly could have caused CK’s injuries. But a reasonable attorney could think that Dr. DeGraw’s willingness to entertain alternative possible explanations for CK’s injuries would actually bolster his credibility with the jury. And Keck sells Dr. DeGraw’s testimony short. Again, Dr. DeGraw testified that the defense’s theory was consistent with CK’s injuries. Dr. DeGraw’s admissions on cross do not establish that calling him as a witness and relying on his testimony violated Strickland.
8
Keck’s next argument points to the testimony of three experts called during his Ginther hearing:1 Dr. Joseph Scheller, a pediatric neurologist; Dr. Roger Haut, a biomechanical engineer; and Dr. Ronald Uscinski, a clinical neurosurgeon. These experts provided additional testimony that Keck posits would have better rebutted the prosecution’s case at trial. Keck argues that, compared to the variety of experts at the Ginther hearing, Dr. DeGraw’s expertise as a child-abuse pediatrician was too narrow. But hiring an expert witness “who, though qualified, was not qualified enough” does not constitute ineffective assistance of counsel. Hinton, 571 U.S. at 275. And there would have been real downsides to calling any of the Ginther experts at trial.
Take Dr. Scheller. At the Ginther hearing, Dr. Scheller attributed CK’s cranial swelling to a pre-existing injury caused by “a short fall of a foot or two” or similar minor trauma. R. 5-15, Ginther Hr’g Tr. Vol. 1, PageID 2621–27. But unlike Dr. DeGraw, Dr. Scheller has only ever testified for the defense. Even worse, Dr. Scheller conceded that his position on the potential causes of infant head trauma represents a fringe minority opinion shared by only five percent of the relevant medical community. And nothing in his testimony explained CK’s fractured ribs or femur. All of this would have severely undercut Dr. Scheller’s credibility.
Or consider Dr. Haut. Dr. Haut is not a medical doctor. He is a biomechanical engineer who has done drop tests on porcelain models. It would be reasonable for trial counsel to think Dr. Haut’s testimony would be more likely to confuse the jury than help it, especially considering he had no explanation for any of CK’s injuries beyond her skull fractures. See Richter, 562 U.S. at
1 A Ginther hearing is an evidentiary hearing available under Michigan law when a criminal “defendant . . . wishes to advance claims that depend on matters not of record . . . .” People v. Ginther, 212 N.W.2d 922, 925 (Mich. 1973). At the Ginther hearing, Keck also called Dr. Douglas Smith, a pathologist, whose testimony explained the types of expert witnesses that are valuable in child-abuse cases.
9
108–09 (highlighting the risk of scientific expert testimony “shift[ing] attention to esoteric matters” or “transform[ing] the case into a battle of the experts”).
Dr. Uscinski fairs worst of all. He testified that CK’s subdural hematoma was most likely caused by an injury during birth. But Dr. Uscinski has provided similar testimony in many criminal cases. As a result, fifty physicians have signed a letter stating that his medical theories are “inaccurate[,] contrary to vast clinical experience[,] and unsupported by any published literature.” R. 5-17, Ginther Hr’g Tr. Vol. 3, PageID 2893. That letter characterized Dr. Uscinski’s theory as “a courtroom diagnosis[,] not a medical diagnosis . . . .” Id. In fact, Dr. Uscinski has been censured by the American Association of Neurological Surgeons for providing testimony that Keck now insists his trial counsel was ineffective for failing to obtain. Declining to call an expert like Dr. Uscinski was not only a reasonable strategic decision—it was a prudent one.
But even if any of those potential expert witnesses would have helped Keck’s case, it is not enough for him to show that “there were any number of hypothetical experts . . . whose insight might possibly have been useful.” Richter, 562 U.S. at 107. Rather, Keck must show that no fair- minded jurist could find the decision not to call these experts consistent with Strickland. He has failed to make that showing.
Keck also faults his trial counsel’s preparation, arguing that Dr. DeGraw’s testimony at the parental-rights-termination hearing “should have raised red flags” and that insufficient research led his counsel to erroneously rely on Dr. DeGraw. Appellant Br. 36–40. But “the duty to investigate does not force defense lawyers to scour the globe on the off chance something will turn up; reasonably diligent counsel may draw a line when they have good reason to think further investigation would be a waste.” Rompilla v. Beard, 545 U.S. 374, 383 (2005). Apart from consulting Dr. DeGraw, a well-credentialed expert in child-abuse pediatrics, trial counsel had
10
preexisting knowledge about abusive head trauma in children because he had previously defended at least two such cases. And he reviewed medical research in preparation for trial in this case. Even if more investigation could have unearthed additional information, Keck’s trial counsel did enough to create a reasonable argument that he satisfied Strickland.
Keck’s final argument on this point relies on People v. Ackley, 870 N.W.2d 858 (Mich.
2015). In that case, the Michigan Supreme Court found counsel ineffective for (1) failing to retain an expert witness; (2) consulting only one expert pretrial, given that expert was completely unsympathetic to the defense theory of the case; and (3) failing to contact any of the potentially more-sympathetic experts to which counsel was referred. Id. at 389–98. Keck frames Ackley as establishing a standard of reasonable conduct for Michigan attorneys. But even if we were to adopt this framing, Ackley is inapposite. Keck’s counsel retained an expert witness—in fact, he retained the only child-abuse pediatrician called at trial. And Dr. DeGraw was sympathetic to Keck’s theory of the case, providing the only medical testimony that supported it. So Ackley does not get Keck anywhere.2 In sum, nothing Keck raises on this point brings him close to meeting his burden under our doubly deferential standard. B. Failure to Provide Dr. DeGraw with Retinal Imaging Before Trial Keck next faults his trial counsel for not providing Dr. DeGraw with CK’s retinal imaging in advance of trial because (1) these images “were an essential part of the [prosecution’s] case”
2 Keck also cites two Supreme Court cases, both of which are far afield of the present case. First, there is Wiggins v. Smith, 539 U.S. 510 (2003), in which counsel was found ineffective for failing to present a mitigation defense in the sentencing portion of a capital murder case. Id. at 534–35. But that is nowhere near the situation here. Second, there is Hinton, where counsel was found ineffective for deciding not to retain a better expert. 571 U.S. at 273–74. The Court held that decision was unreasonable because it was borne out of the mistaken belief that there was insufficient funding available, rather than any strategic consideration. Id. Here, Keck’s trial counsel made the strategic choice to retain Dr. DeGraw. The record does not suggest costs had anything to do with that decision.
11
and (2) the fact that Dr. DeGraw had not seen them before trial caused him to defer to the prosecution’s expert in pediatric ophthalmology when questioned about the images on cross. Appellant Br. 38–39. The Michigan Court of Appeals did not find this argument compelling, Keck, 2022 WL 128582, at *11, and neither do we.
Even though Dr. DeGraw did not have pretrial access to the retinal imaging, he had pretrial access to CK’s medical records, which included the findings of her treating ophthalmologist. Keck does not say how pretrial access to the retinal imaging would have changed Dr. DeGraw’s testimony. On that point alone, Keck fails to meet his burden.
Keck’s real gripe seems to be with Dr. DeGraw for deferring to the pediatric ophthalmologist who treated CK at the hospital. But that deference had nothing to do with a lack of access to the retinal imaging. Dr. DeGraw testified to the supreme value of visual, in-person observations, and agreed that he “would defer to [the treating ophthalmologist’s] opinion because she did visually see the child.” R. 9-17, 8/10/18 Trial Tr., PageID 4462. So trial counsel’s failure to provide Dr. DeGraw with the retinal imaging did not cause the testimony about which Keck now complains.
Finally, Keck ignores the fact that Dr. DeGraw’s testimony about CK’s retinal hemorrhaging was beneficial to the defense. Dr. DeGraw testified that her preretinal hemorrhages were not “terribly significant,” and that none of the hemorrhages were “all that severe.” Id. at PageID 4464. And he went on to disagree, at length, with the conclusions drawn by the prosecution’s medical witness about CK’s retinal hemorrhaging. So even if it was unreasonable not to provide Dr. DeGraw with the retinal imaging before trial, nothing suggests Keck was prejudiced. This claim cannot clear the bar of double deference.
12
C. Insufficient Impeachment Keck’s final argument on appeal concerns trial counsel’s failure to produce impeachment evidence while cross-examining prosecution witness Dr. Elana Gianfermi. This expert testified that CK’s retinal hemorrhages were less than a week old when she went to the hospital. On cross, Keck’s trial counsel asked Dr. Gianfermi if she remembered testifying in a previous hearing that the injuries could have been up to two weeks old. After Dr. Gianfermi answered no, trial counsel failed to counter with evidence of her prior inconsistent statements. The Michigan Court of Appeals found that this deficient cross-examination was objectively unreasonable but that Keck suffered no prejudice because the testimony concerned an issue of limited importance compared to the overwhelming evidence the prosecution provided. Keck, 2022 WL 128582, at *6. We do not think the state court’s determination that Keck failed to show prejudice was unreasonable.
The standard for showing prejudice under Strickland is high. “Representation is constitutionally ineffective only if it ‘so undermined the proper functioning of the adversarial process’ that the defendant was denied a fair trial.” Richter, 562 U.S. at 110 (quoting Strickland, 466 U.S. at 686). It is not enough for Keck to show the mere possibility that better impeachment would have returned a favorable jury verdict; the likelihood of a different outcome must be “substantial, not just conceivable.” Id. at 112. A fair-minded jurist could conclude Keck has not shown that here.
Keck argues that if CK’s retinal hemorrhaging could have occurred as early as two weeks prior to her hospital visit, it would have allowed him to more plausibly allege that her injuries were caused by the actions of her siblings rather than Keck. There are several problems with this argument.
13
The first is that the retinal hemorrhaging was not the only subject of timeline testimony at trial. CK’s hospital diagnoses included extensive head injuries, including cranial fractures and a subdural hematoma. Prosecution witness Dr. Deniz Altinok, a pediatric neuroradiologist at the hospital, estimated her head injuries were less than a week old when she arrived. Better impeachment of Dr. Gianfermi would not have rebutted this testimony establishing a shorter timeline for CK’s other head injuries.
The second and most formidable stumbling block for Keck’s argument is that the evidence against him at trial was overwhelming. The prosecution called several medical doctors who were involved with CK’s treatment at the hospital. All testified that her injuries were most likely caused by nonaccidental head trauma—i.e., abuse—and that the defense’s theory of how she was injured was implausible. The jury also heard extensive testimony about Keck’s previous conviction for the second-degree murder of his infant child under very similar circumstances. Keck not only needs to show that a more thorough impeachment of Dr. Gianfermi would have overcome such a mountain of evidence to deliver him an acquittal—he also needs to show that no fair-minded jurist could disagree with him on that point. He has not done so.
IV.
For the foregoing reasons, we AFFIRM the district court’s denial of Keck’s habeas petition.