Louis Chandler v. Mike Brown

126 F.4th 1178

• Court: Court of Appeals for the Sixth Circuit
• Decided: January 24, 2025
• Docket: 23-1270
• Status: Published

Opinion

RECOMMENDED FOR PUBLICATION Pursuant to Sixth Circuit I.O.P. 32.1(b) File Name: 25a0017p.06

UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT

┐ LOUIS CHANDLER, │ Petitioner-Appellant, │ > No. 23-1270 │ v. │ │ MIKE BROWN, Warden, │ Respondent-Appellee. │ ┘

Appeal from the United States District Court for the Western District of Michigan at Marquette. No. 2:19-cv-00263—Paul Lewis Maloney, District Judge.

Argued: April 30, 2024

Decided and Filed: January 24, 2025

Before: WHITE, STRANCH, and DAVIS, Circuit Judges.

_________________

COUNSEL

ARGUED: Matthew A. Monahan, STATE APPELLATE DEFENDER OFFICE, Detroit, Michigan, for Appellant. Jared D. Schultz, OFFICE OF THE MICHIGAN ATTORNEY GENERAL, Lansing, Michigan, for Appellee. ON BRIEF: Matthew A. Monahan, STATE APPELLATE DEFENDER OFFICE, Detroit, Michigan, for Appellant. Jared D. Schultz, OFFICE OF THE MICHIGAN ATTORNEY GENERAL, Lansing, Michigan, for Appellee. _________________

OPINION _________________

HELENE N. WHITE, Circuit Judge. Petitioner-Appellant Louis Chandler, a Michigan prisoner, is serving concurrent terms of twenty-five to seventy-five years in prison for two convictions of first-degree criminal sexual conduct. Mich. Comp. Laws § 750.520b(2)(b). No. 23-1270 Chandler v. Brown Page 2

After exhausting his state-court appeals, Chandler filed a petition for habeas corpus pursuant to 28 U.S.C. § 2254, claiming that the trial court infringed his right to present a complete defense. The district court denied the petition, and Chandler appeals. We REVERSE, conditionally GRANT Chandler’s habeas corpus petition, and REMAND to the district court.

I. Background

A. Factual History

For twelve years, Chandler and his wife, Darlene Chandler (“Darlene”), cared for more than twenty foster children without incident. In 2010, the Chandlers decided to foster A.C., an eight-year-old girl.1 Because a prior Child Protective Services (“CPS”) report had concluded that A.C. had a “history of false allegations,” R. 9-10, PID 728, the foster-care agency warned the Chandlers to “watch” A.C. closely, R. 9-5, PID 382.

Roughly three months after the initial foster placement, the Chandlers informed A.C. that they intended to adopt her. Days later, A.C. told Darlene that Chandler had touched her inappropriately. Darlene was confused because, shortly after reporting this incident to Darlene, A.C. “laugh[ed] around the room . . . and climbed on [Chandler’s] lap.” R. 9-5, PID 380. Still, Darlene called the adoption agency to report A.C.’s allegations, and Chandler voluntarily left the home to give law enforcement, CPS, and the foster-care agency time to investigate.

A.C. told investigators that Chandler sexually abused her every day, and sometimes multiple times per day, both inside the home and at a local movie theater.2 The foster-care agency concluded that it had serious doubts regarding A.C.’s assertions, which were compounded by A.C.’s history of false allegations. The prosecutor also declined to charge Chandler at the time. However, the CPS investigator assigned to the case, Jennifer Schmidt, concluded that the allegations were substantiated under a preponderance of the evidence standard. Schmidt had not read A.C.’s file, which detailed her history of false allegations, but

1 The State’s brief uses the initials “A.H.” instead of “A.C.,” perhaps because the child’s last name changed after adoption. For consistency, we use “A.C.” 2 CPS also interviewed other foster children who were staying with the Chandlers at the time, but the interviews did not turn up anything of note. No. 23-1270 Chandler v. Brown Page 3

another foster-care worker had told Schmidt that A.C. had made allegations of physical abuse in the past. After Schmidt’s investigation, CPS removed A.C. from the Chandlers’ home, and a new family eventually adopted her.3 In the following years, a dispute arose within the Chandler family. During the dispute, the Chandlers’ son—Louis Chandler Jr. (“Lou Jr.”)—allegedly threatened to kill his mother, and Darlene testified that she was afraid of her son.

In 2014, Lou Jr.’s eight-year-old stepdaughter, Z.B., told her mother, Stephanie Chandler (“Stephanie”), that Chandler had touched her inappropriately. Lou Jr. and Stephanie filed a police report, which triggered a second criminal investigation, and told police that they knew of two other victims: Stephanie’s best friend, Josefina Harden, and Norma Chandler (“Norma”), Chandler’s sister.

Police also re-interviewed A.C., who was then twelve years old. Detective Svoboda, who authored a report following the interview, testified that A.C.’s account was inconsistent with the one she gave previously. In 2010, A.C. alleged that Chandler abused her every day, both at home and at the movie theater. But in her 2014 interview, A.C. told the detective that Chandler abused her four times total, including once after a church party, but never at a movie theater.

Police then obtained a search warrant for Chandler’s computer, which yielded written stories of a sexual nature, including one containing child sexual abuse. Several other individuals used the computer in addition to Chandler, however, and Darlene testified that she once caught her adopted teenage son looking at “real[ly] bad” pornography on the computer. R. 9-5, PID 380. Darlene also testified that pornography would pop up randomly on its own when she used the computer. In 2015, the Kent County prosecutor charged Chandler with four counts of first- degree criminal sexual conduct stemming from the alleged sexual abuse of A.C., but he dismissed two of the charges before trial.

3 CPS also filed a petition requesting jurisdiction of Austin, the Chandlers’ adopted son. R. 9-4, PID 331. Although no record document discloses whether the petition was successful, Austin still lived with the Chandlers at the time of trial. No. 23-1270 Chandler v. Brown Page 4

B. Chandler’s Motions

The case had an “accelerated” timeline. People v. Chandler, No. 329605, 2017 WL 6502801, at *2 (Mich. Ct. App. Dec. 19, 2017). On April 10, 2015, the court appointed Jonathan Schildgen as Chandler’s public defender. Chandler entered a not-guilty plea on April 22, and discovery began a day later. Schildgen received the first batch of discovery documents in mid- May. On May 26, the court set the case for trial on July 6—just seventy-five days after Chandler entered his not-guilty plea.

As Schildgen went through the initial document disclosure, he found the CPS report concluding that A.C. had a “history of making false allegations” against foster parents. R. 9-3, PID 256. Based on this new information, Chandler successfully requested funds on June 11 to hire an expert on child-sexual-abuse allegations.

On June 16, Chandler moved to compel additional discovery, seeking “all Kent County DHS and CPS reports for [A.C.] containing and regarding references to prior false and prior unsubstantiated sexual abuse allegations.” R. 9-10, PID 519. The court held a hearing on June 26 and denied the motion, concluding that Chandler was only entitled to records related to A.C.’s allegations against him and that the prosecution did not need to disclose any evidence of A.C.’s allegations against others.4

At the same hearing, Schildgen also requested an adjournment of the trial date, explaining that he still needed to obtain additional evidence and that his expert needed time to review the records. The court denied the request. On June 29, Chandler moved for reconsideration, arguing that the expedited timeline deprived him of the right to present a defense and made it impossible for Schildgen to represent him effectively. The court denied the request.

On the morning of the July 6 trial, Chandler again moved for an adjournment. Schildgen told the court that he had received additional discovery from the foster-care agency just a week earlier, including a report raising “serious doubts” as to whether sexual abuse occurred. Id. at

4 The court told Schildgen to “go out and talk to [the CPS workers], you can sit down and meet with them. . . . [T]hat burden is on your shoulders.” R. 9-10, PID 520. No. 23-1270 Chandler v. Brown Page 5

521. The new documents included the names of other foster parents who A.C. had falsely accused of misconduct. Schildgen argued that, because he received the records only a week earlier, he needed additional time to go through them, to provide them to the expert, and to prepare for trial.

Schildgen also found a probate case involving A.C., in which Judge Kathleen Feeney issued an order with factual findings useful to Chandler. Judge Feeney’s order concluded that A.C. had issues with lying and making false allegations, motivated in large part by her desire to be removed from foster care and returned to her birth parents. Schildgen told the trial court that Judge Feeney had agreed to make the records available for Chandler’s proceedings, but she could only do so after returning from a two-week vacation. Schildgen thus explained that delaying trial would give him time to pick up Judge Feeney’s records, go through them, and make them available to the expert.

The trial court rejected Chandler’s arguments, concluding that any evidence of A.C.’s prior allegations was inadmissible because it was “extrinsic.” R. 9-3, PID 258. Chandler could only raise the topic through cross-examination of A.C., but he would be “stuck with her answers.” Id. The court also barred Chandler’s expert from testifying at trial because Chandler had not complied with the court rule requiring disclosure of an expert’s name “no later than 28 days before trial.”5 See Mich. Ct. R. 6.201(A)(1). Finally, the trial court found no legitimate reason to postpone trial and denied Chandler’s motion.

The trial court also barred Chandler’s remaining proposed witnesses from testifying— more than twenty of them—because their names were not disclosed at least ten days before trial. Chandler argued that, because he was still receiving discovery until the day of trial, it would have been impossible to comply with this rule. Among those that Chandler told the trial court he intended to call to testify were: (1) three of A.C.’s previous foster families—the Hamblins, Nickersons, and Lamberts—who would have testified regarding A.C.’s character and her history of false allegations, including that a foster parent heard A.C. tell her therapist that “it’s fun to

5 The court authorized funding for the expert twenty-five days before trial, so it was impossible for Chandler to comply with the rule. Still, Chandler failed to disclose the expert’s name until the day of trial, which he argued was because the expert had insufficient time to review the relevant evidence and create a report. No. 23-1270 Chandler v. Brown Page 6

lie.” (2) Amanda Fraly, who would have testified that Lou Jr. pressured her to make a false allegation against Chandler before he contacted police; and (3) five foster-care-agency employees, who would have testified about A.C.’s false allegations and Chandler’s twelve years as a “model caregiver.” R. 9-5, PID 348. Chandler again tried to call witnesses on the second day of trial, arguing that he was denied his “constitutional right to present a defense,” R. 9-4, PID 340, but the court refused to permit any witnesses not already “endorsed by the prosecution,” id. at 339. In sum, the court’s orders limited Chandler’s time to prepare for trial, stymied his discovery, and prevented him from calling any witnesses.

C. Trial Testimony

A.C. testified at Chandler’s trial and provided inconsistent testimony. At first, A.C. testified that (1) she was unable to recognize Chandler or Darlene, (2) she did not remember making any allegations against Chandler, (3) she did not remember any CPS investigation, (4) she did not remember any foster placement before her current home, and (5) she did not remember any time when she was touched inappropriately. Moments later, A.C.’s testimony changed. She testified that Chandler sexually abused her twice—both times in the Chandlers’ home. A.C.’s testimony contradicted her 2010 allegations that Chandler abused her daily, both at home and elsewhere, and the account she gave Svoboda four months earlier that Chandler abused her four times, including once after a church party.

When asked about the inconsistent accounts on cross-examination, A.C. denied alleging in 2010 that Chandler’s abuse happened daily, both inside and outside the home, and did not remember making the inconsistent statements to Svoboda months earlier. A.C. also reiterated that Chandler’s abuse occurred twice in total. Schildgen attempted to challenge A.C.’s honesty by questioning her about her history of stealing from retail establishments, but the court cut off this line of questioning.

Finally, Schildgen asked A.C. about a lengthy list of allegations that she had purportedly made in the past. Schildgen learned of these prior allegations after obtaining CPS and foster- care-agency records in the days before trial. The records included allegations that (1) a foster family’s dog had attacked her, (2) a foster parent abused her by swinging her around by her No. 23-1270 Chandler v. Brown Page 7

ponytail, (3) a foster parent refused to give her clothing, shoes, or bedding, (4) a foster parent hit her with a wooden spoon, (5) a cousin abused her, (6) a daycare provider had hit her, (7) a foster parent sexually abused her, (8) a foster parent made her eat soap, and (9) a foster-care worker raped her. R. 9-4, PID 321–24; R. 9-7, PID 413; R. 9-10, PID 619. In her testimony, A.C. admitted to making only two of these allegations, but she testified that both were true.6 For the others, A.C. either denied or did not remember making the allegations. Because the court had barred the defense from calling witnesses and citing the CPS records directly, there was no evidence at trial to contradict A.C.’s claim that she never made false allegations.

The prosecution then called three other witnesses who testified that Chandler had touched them inappropriately in the past.

Z.B. first testified that, over the span of two years, Chandler sexually abused her once or twice on each day they spent together. However, in a prior proceeding, Z.B. had given contradictory testimony. On cross-examination, Z.B. testified that she did not remember making the inconsistent statements.7

Then, Norma testified that Chandler had molested her nearly fifty years earlier when they were both children. R. 9-5, PID 350. At the time of the alleged abuse, Norma was eight to ten years old and Chandler was “eleven or twelve.” R. 9-5, PID 351.

Finally, Josefina Harden, Stephanie’s “best friend,” described an incident that occurred roughly seventeen years earlier. R. 9-4, PID 338. Harden testified that when she was about ten years old, she went to the Chandlers’ home for dinner. Because Harden had just been playing in a sprinkler, she was wearing a bathing suit and had sand all over her body. Chandler took Harden to the bathroom to clean the sand off her body and allegedly used his hands to brush off the sand between her legs, including a “quick swipe” to clean the sand “just outside of . . . the genital area.” Id. at 337.

6 Of these, the only substantiated allegation was that a foster parent put a dab of soap in A.C.’s mouth after A.C. called the foster parent “a fucking bitch.” R. 9-10, PID 812. 7 Evidence that Chandler passed a polygraph test concerning Z.B.’s allegations was disallowed. No. 23-1270 Chandler v. Brown Page 8

The prosecution also called Thomas Cottrell for expert testimony. Cottrell was a professor at Western Michigan University and the executive of a company providing counseling to survivors of sexual assault. Cottrell testified that children often wait to disclose sexual abuse, and roughly half of survivors wait until adulthood. He explained that late disclosures generally occur for one of three reasons: the victim (1) realizes in adulthood that abuse occurred, (2) disassociates from the memory due to trauma, or (3) believes that the costs of disclosure outweigh the benefits. Although victims’ memories may degrade over time, Cottrell testified that memories can “become enriched” as an individual discusses them repeatedly. Id. at 370. Cottrell also testified that, because children have difficulty with chronology, they can “confound multiple occurrences of abuse,” making it difficult to distinguish them. Id.

On cross-examination, Schildgen asked about false allegations of sexual abuse, and Cottrell testified that young children could come to believe lies by repeating them. Schildgen also asked about Reactive Attachment Disorder (“RAD”), a condition that occurs in children who have been deprived of a relationship with their parents at a young age. Cottrell explained that children diagnosed with RAD resist relationships, have difficulty trusting others, and are not “tuned into the consequences of their choices.” Id. at 372. Schildgen asked about an incident where A.C. allegedly told a therapist that she enjoyed lying, which Cottrell said was “not atypical of [a RAD] diagnosis.” Id.

The jury convicted Chandler on two counts of first-degree criminal sexual assault, and the trial court sentenced him to concurrent sentences of twenty-five to seventy-five years in prison. During sentencing, the judge said that he was “absolutely convinced that [Chandler is] a pedophile” and that he did not “ever want [Chandler] out of prison.” R. 9-7, PID 420.

D. Remand

Chandler appealed, arguing that the trial court denied him his right to present a complete defense and incorrectly applied the state’s evidentiary and procedural rules. The Michigan Court of Appeals remanded the case for an evidentiary hearing, which was held on August 24, 2017.

At the hearing, Schildgen testified that he had intended to call A.C.’s former foster parents and caseworkers to discuss specific instances of A.C.’s false allegations. Schildgen had No. 23-1270 Chandler v. Brown Page 9

also planned to call an expert witness to testify about Reactive Attachment Disorder, proper interviewing techniques, and false accusations of sexual assault. Schildgen argued that, because he was unable to call any witnesses, the jury had only A.C.’s answers without any consideration of evidence impeaching her credibility.

Sandy and Randy Hamblin, the foster parents who cared for A.C. before the Chandlers, also testified. Sandy testified that although A.C. had difficulty attaching to her, she was comfortable with the men in her family. A.C. “wanted to sit on their lap[s] all the time,” but Sandy discouraged this behavior because she had heard about A.C.’s prior allegations. R. 9-10, PID 763. Sandy also detailed several allegations that A.C. made against the Hamblins, which were later determined by CPS to be “unfounded.” Id. at 764. As with Chandler, A.C. made allegations against the Hamblins days after she learned that they intended to adopt her. Randy corroborated Sandy’s testimony and, when asked his opinion of A.C.’s character, testified that she “could not tell the truth.” Id. at 766.

Jeff Kieliszewski, Chandler’s expert witness, then testified about “confabulation,” a type of memory error where an individual produces fabricated, distorted, or misrepresented memories. Id. at 768. He explained how a child in A.C.’s position could come to believe things that were not true, particularly after repeating the story many times, id., and that, because memory degrades over time, it was a “red flag” that A.C. added additional details to her story years after it allegedly occurred. Id. at 770. Kieliszewski also discussed the ways in which the investigators failed to follow standard interviewing procedures with A.C. For example, because a parent may influence a child’s answers or pressure them to say something untrue, the guidelines for child forensic interviews “highly discourage” the presence of a support person. Id. at 769. But A.C.’s adoptive mother was allowed to sit next to her for the interview, which the defense would have argued made the answers unreliable. Id. The guidelines also recommend recording interviews as a “best practice,” but the detectives did not do so with A.C. Id. Instead, the detectives wrote a one-and-a-half-page report, which the expert said was “quite short” for a ninety-minute interview. Id. After reviewing A.C.’s record and the investigators’ interviewing techniques, Kieliszewski concluded that there was a “substantial possibility of a false allegation report of sexual abuse.” Id. at 774. No. 23-1270 Chandler v. Brown Page 10

After the testimony concluded, the prosecution argued that a new trial was unnecessary because (a) the documents Schildgen received on the eve of trial had no new information compared to the documents he received earlier, (b) Chandler successfully made his core arguments by cross-examining the government’s witnesses, and (c) the evidence and testimony that Chandler sought to present at trial was inadmissible.

The trial court agreed with the prosecution and rejected Chandler’s request for a new trial, concluding that all the evidence Schildgen had sought to introduce was inadmissible, except for the expert’s testimony regarding proper interviewing procedures. It also determined that, to the extent any testimony was wrongly excluded, the error did not prejudice Chandler because the “evidence was clearly overwhelming.” Id. at 779.

E. Appeal

Chandler’s case returned to the Michigan Court of Appeals, where he again argued that the trial court had violated the state’s trial rules and denied him the right to present a complete defense. In a brief footnote, the court rejected Chandler’s constitutional claim, concluding that Chandler had a meaningful opportunity to present a complete defense because (1) he was represented by counsel at trial and (2) his counsel could argue through cross-examination that A.C. fabricated the allegations. Chandler, 2017 WL 6502801, at *4 n.3.

However, the court of appeals found Chandler’s evidentiary and procedural claims meritorious. It first concluded that the trial court abused its discretion by denying Chandler’s repeated requests for an adjournment without any “reasonable or principled basis.” Id. at *3. The trial court further abused its discretion by barring all of Chandler’s lay and expert witnesses from testifying because it was a disproportionally “extreme sanction” for a minor procedural violation. Id. Finally, because the trial court “employed the wrong framework when considering the admissibility of extrinsic evidence,” it abused its discretion by excluding evidence of A.C.’s prior false allegations. Id. at *4. Instead, the trial court should have considered admitting the evidence under Michigan Rule of Evidence 404(b), id., which allows admitting extrinsic evidence “for a[] purpose, such as proving motive, opportunity, intent, preparation, scheme, plan, No. 23-1270 Chandler v. Brown Page 11

or system in doing an act, knowledge, identity, absence of mistake, or lack of accident,” Mich. R. Evid. 404(b). Thus, the jury should have heard evidence of A.C.’s prior false allegations.

Despite the trial court’s multiple abuses of discretion, the court of appeals affirmed Chandler’s conviction under the state’s forgiving test for non-constitutional errors, which allows a court to overturn a conviction only if “it affirmatively appears that it is more probable than not that the error was outcome determinative.” Chandler, 2017 WL 6502801, at *4 (quoting People v. King, 824 N.W.2d 258, 262 (Mich. Ct. App. 2012)). The court concluded that reversal was unwarranted because (1) Chandler challenged A.C.’s credibility during cross-examination by asking her about her prior inconsistent statements, (2) Chandler’s expert would have discussed the same topics as the prosecution’s expert, and (3) the testimony of Chandler’s other instances of alleged sexual misconduct “bolstered the victim’s credibility and supported a propensity inference.” Id. at *5.

F. District Court

On December 30, 2019, Chandler filed this petition for habeas corpus in federal district court, arguing that he was denied several due process rights—including the right to present a complete defense, to call witnesses on his own behalf, and to a fair trial. On February 3, 2023, the magistrate issued a report and recommendation to deny Chandler’s petition. Chandler objected to the report and recommendation, but the district court overruled the objections and adopted the recommendation. Although the district court concluded that Chandler could not establish that the state court unreasonably applied clearly established law, it granted a certificate of appealability because a reasonable jurist could disagree. R. 15, PID 1021–22.

Chandler appeals.

II. Constitutional Claim

A. Standard of Review

“AEDPA requires habeas petitioners to exhaust their claims in state court before turning to a federal court for relief.” Stermer v. Warren, 959 F.3d 704, 720 (6th Cir. 2020). A state court’s resolution of a claim on the merits receives deference in federal habeas proceedings. Id. No. 23-1270 Chandler v. Brown Page 12

Accordingly, a federal court may grant relief to a petitioner only when a state court’s decision is “(1) . . . contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States; or (2) . . . based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” Id. at 720–21 (alterations in original) (quoting 28 U.S.C. § 2254(d)). “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 (2011) (quoting Yarborough v. Alvarado, 541 U.S. 652, 664 (2004)).

An “unreasonable application of federal law is different from an incorrect application of federal law.” Williams v. Taylor, 529 U.S. 362, 410 (2000). A federal court may grant the writ under the “contrary to” clause “if the state court arrives at a conclusion opposite to that reached by th[e] Court on a question of law or if the state court decides a case differently than th[e] Court has on a set of materially indistinguishable facts.” Id. at 412–13. Under the “unreasonable application” clause, a federal court may grant the writ “if the state court identifies the correct governing legal principle from th[e] Court’s decisions but unreasonably applies that principle to the facts of the prisoner’s case.” Id. at 413. In determining whether the state court applied a rule unreasonably, a federal court must consider the rule’s specificity. Yarborough, 541 U.S. at 664. “The more general the rule, the more leeway courts have in reaching outcomes in case-by-case determinations.” Id.

“In a habeas appeal, we review questions of law de novo, including the ultimate decision to grant or deny the petition.” Stermer, 959 F.3d at 720. Absent an evidentiary hearing, the district court’s factual findings are also reviewed de novo. Id. We take care to consider the “entire record” in the case, including both the trial and remand hearing transcripts. Mays v. Hines, 592 U.S. 385, 392 (2021); see also Adams v. Holland, 330 F.3d 398, 406 (6th Cir. 2003) (holding that federal courts must consider habeas petitions “in light of the full record”).

B. Right to Present a Complete Defense

The Constitution guarantees criminal defendants “a meaningful opportunity to present a complete defense.” California v. Trombetta, 467 U.S. 479, 485 (1984). This right is rooted in No. 23-1270 Chandler v. Brown Page 13

several constitutional provisions, including the Fourteenth Amendment’s Due Process Clause and the Sixth Amendment’s various trial rights. Crane v. Kentucky, 476 U.S. 683, 690 (1986); see also Strickland v. Washington, 466 U.S. 668, 684–85 (1984) (“The Constitution guarantees a fair trial through the Due Process Clauses, but it defines the basic elements of a fair trial largely through the several provisions of the Sixth Amendment.”); In re Oliver, 333 U.S. 257, 273 (1948) (“A person’s right to reasonable notice of a charge against him, and an opportunity to be heard in his defense—a right to his day in court—are basic in our system of jurisprudence; and these rights include, as a minimum, a right to examine the witnesses against him, to offer testimony, and to be represented by counsel.”).

1. Key Precedents

Because Chandler must show that the state court contradicted or unreasonably applied clearly established federal law as established by the Supreme Court, we begin by surveying the key precedents concerning the right to present a complete defense. We survey both Supreme Court cases defining the right to present a complete defense and our own cases applying this clearly established federal law under the AEDPA standard.

Washington v. Texas, 388 U.S. 14 (1967)

Jackie Washington was charged with murder. Id. at 15. The only eyewitness was Charles Fuller, an alleged accomplice who had already been tried and convicted for the same crime. Id. at 16. Although Fuller shot the victim, not Washington, he was barred from testifying due to a Texas evidentiary rule that prohibited accomplices from testifying on behalf of the defense. Id. at 16–17. The Supreme Court explained that Texas’s rule “arbitrarily denied [Washington] the right to put on the stand a witness who was physically and mentally capable of testifying to events that he had personally observed, and whose testimony would have been relevant and material to the defense.” Id. at 23. Because criminal defendants have the right to present their own “version of the facts,” including the “right to present [their] own witnesses to establish a defense,” the Court held that Texas’s rule unconstitutionally denied Washington a fair trial. Id. at 19. No. 23-1270 Chandler v. Brown Page 14

Chambers v. Mississippi, 410 U.S. 284 (1973)

Leon Chambers was charged with murder, but no physical evidence tied him to the crime. Id. at 287, 289. Another man, Gable McDonald, confessed to the murder on four occasions and was arrested, but he later repudiated the confession. Id. at 287–88. At trial, Chambers tried to show that it was McDonald who had committed the murder. Id. at 289. One witness testified that he saw McDonald shoot the victim, and another witness testified that he saw McDonald immediately after the shooting with a gun in his hand. Id. A third witness testified that he was with Chambers during the shooting and did not see him holding a firearm. Id. Chambers called McDonald to testify, but he again repudiated his confession. Id. at 291. Due to a state rule of evidence, the trial court prevented Chambers from treating McDonald as an adverse witness and from impeaching his testimony. Id. Chambers was also prohibited from calling other witnesses to testify that they heard McDonald’s confession. Id. at 292.

The Court overturned Chambers’s conviction, explaining that by excluding evidence “critical to Chambers’ defense,” the trial court “denied him a trial in accord with traditional and fundamental standards of due process.” Id. at 302. (“Few rights are more fundamental than that of an accused to present witnesses in his own defense.”). Notably, Chambers had competent counsel and called several strong eyewitnesses to support his alternate-suspect theory—including one who saw McDonald shoot the victim and another who saw him holding a gun—but the Court still held that Chambers was entitled to a new trial because he was denied the opportunity to present a complete defense.

Ungar v. Sarafite, 376 U.S. 575 (1964)

Sidney Ungar was charged with criminal contempt after giving witness testimony, and the court scheduled his contempt hearing roughly three weeks later. Id. at 580–81. On the day of the hearing, Ungar requested a continuance because he had hired a new lawyer five days earlier who was unfamiliar with the facts of the case. Id. at 590. The trial court rejected Ungar’s request. Id. On appeal, the Supreme Court held that the denial of a continuance did not violate due process because the facts of the case were very simple—implicating only a single sentence of Ungar’s testimony. Id. However, the court explained that, in a more complicated case, “a No. 23-1270 Chandler v. Brown Page 15

myopic insistence upon expeditiousness in the face of a justifiable request for delay can render the right to defend with counsel an empty formality.” Id. at 589.8 Accordingly, a court may not arbitrarily deny a defendant’s request to delay trial when it impedes the right to a fair trial.

O’Neal v. Balcarcel, 933 F.3d 618 (6th Cir. 2019)

Tyson O’Neal was charged with second-degree murder. Id. at 620. At trial, O’Neal attempted to argue that another man, Parish Hickman, shot the victim. Id. However, the trial court excluded two statements supporting defendant’s theory: (1) a jailhouse confession that Hickman made to another inmate and (2) the victim’s dying declaration to a police officer identifying Hickman as the shooter. Id. at 622. The trial court excluded both statements—the first because it was not timely disclosed to the prosecution and the second because it was hearsay. Id. Because the excluded evidence was critical to O’Neal’s defense, a panel of this court granted habeas relief on the grounds that the trial court acted contrary to the Supreme Court’s precedent in Chambers and this violation was not harmless under the “grave doubt” standard set forth in Brecht v. Abrahamson, 507 U.S. 619 (1993). Id. at 628.

Ferensic v. Birkett, 501 F.3d 469 (6th Cir. 2007)

Robert Ferensic was tried on charges related to armed robbery and home invasion. Id. at 470. The entirety of the direct evidence against Ferensic consisted of two eyewitness identifications made by the victimized couple. Id. Ferensic intended to call two witnesses in his defense, but the trial court barred them from testifying. Id. at 471. The first would have testified that he saw the two culprits right before the crime and that Ferensic did not resemble either one. Id. The second witness, an expert on eyewitness identifications, would have testified about the potential unreliability of victim identifications. Id. at 471–72. Because the eyewitnesses’ identifications were the strongest evidence against Ferensic, the panel concluded

8 Chandler v. Fretag is one such case where the denial of a continuance violated due process. 348 U.S. 3 (1954). There, a defendant originally waived his right to counsel, but then changed his mind once he learned he would be charged as a habitual offender. Id. at 4–5. The defendant requested a continuance so that he would have time to find new counsel, but the trial court denied the request. Id. The Court held that the denial violated due process. Id. at 10 (“A necessary corollary [of the right to counsel] is that a defendant must be given a reasonable opportunity to employ and consult with counsel; otherwise, the right to be heard by counsel would be of little worth.”). No. 23-1270 Chandler v. Brown Page 16

that the most important issue at trial was whether the identifications were accurate. Id. at 475– 80. But the trial court’s exclusions significantly undermined Ferensic’s ability to cast doubt on the identifications. Id. Relying on Chambers, Washington, and other relevant Supreme Court precedents, the panel concluded that the state court acted contrary to clearly established federal law in denying Ferensic a meaningful opportunity to present a complete defense. Id.

2. Chandler’s Defense

Other than A.C.’s testimony, there was no direct evidence supporting the charges against Chandler. Accordingly, the decisive issue at trial was the reliability of A.C.’s testimony, and Chandler’s defense hinged on challenging her credibility. Chandler attempted to present a complete defense. Before trial, Chandler repeatedly requested an adjournment—both to obtain crucial documents, like Judge Feeney’s order discussing A.C.’s credibility, and to give his counsel sufficient time to read through last-minute discovery. Chandler intended to call more than twenty witnesses at trial—including A.C.’s previous foster parents and foster-care workers—who would have discussed A.C.’s history of false allegations. The defense also secured an expert, who would have testified about false allegations of child sexual abuse and explained how the investigators’ interviewing methods may have been faulty. However, the trial court hampered Chandler’s efforts on all fronts, preventing him from challenging A.C.’s credibility in any meaningful way. When A.C. denied ever having made false allegation on cross-examination, the trial court told Chandler he was “stuck” with her answers and could not challenge them with contradicting evidence. R. 9-3, PID 258. Chandler repeatedly asked the trial court to reconsider the exclusion of his witnesses, but the court refused to permit any witnesses not “endorsed by the prosecution,” R. 9-4, PID 339.

States have “broad latitude under the Constitution to establish rules excluding evidence from criminal trials,” United States v. Scheffer, 523 U.S. 303, 308 (1998), but the state appellate court here concluded that the trial court violated the state’s rules by improperly excluding evidence, barring witness testimony, and refusing to delay the trial, Chandler, 2017 WL 6502801, at *3–4. Accordingly, we begin with the premise that the trial court’s rulings were not justified by any of the state’s evidentiary or procedural rules. No. 23-1270 Chandler v. Brown Page 17

The State argues that we should afford no deference to a state court’s “pro-petitioner” resolution of an issue, citing Daniels v. Lafler, 501 F.3d 735, 740 (6th Cir. 2007). The State misreads Daniels, a case in which the state court considered the defendant’s constitutional challenge hypothetically. Id. at 739–40. The state court did not decide the constitutional issue on the merits, nor did it make any evidentiary rulings that were decisive in the case. Id. A panel of this court reviewed the constitutional issue de novo because AEDPA deference “applies only to claims ‘adjudicated on the merits in State court proceedings,’ and the standard of review it mandates depends on an assessment of an actual decision made by the state court.” Id. at 740 (quoting Eddleman v. McKee, 471 F.3d 576, 583 n.3 (6th Cir. 2006)). By contrast, the state court’s holdings in Chandler’s case were not hypothetical—they were adjudications on the merits. Accordingly, we accept the state court’s holdings that the trial court repeatedly abused its discretion by violating the state’s evidentiary and procedural rules.

But even if the trial court’s actions were sanctioned by state rules, at least some of its rulings would not have withstood constitutional scrutiny. A trial court “abridge[s] an accused’s right to present a defense” when its exclusion of evidence is “‘arbitrary’ or ‘disproportionate to the purposes . . . serve[d].’” Scheffer, 523 U.S. at 308 (quoting Rock v. Arkansas, 483 U.S. 44, 56 (1987)). Exclusion of evidence is “unconstitutionally arbitrary or disproportionate only where it has infringed upon a weighty interest of the accused.” Id. Accordingly, reviewing courts are required to apply a proportionality test, carefully balancing the state’s interests in exclusion against the “weighty interests” of the defendant. Ferensic, 501 F.3d at 476–77 (quoting Scheffer, 523 U.S. at 308). However, excluding evidence is an excessive penalty in most cases where there is a discovery violation. See Michigan v. Lucas, 500 U.S. 145, 152 (1991) (holding that, “in most cases,” exclusion would be unnecessary because “alternative sanctions would be ‘adequate and appropriate’” (quoting Taylor v. Illinois, 484 U.S. 400, 413 (1988)). Accordingly, excluding evidence in response to a discovery violation is appropriate “only [in] egregious violations involving, for example, ‘willful misconduct’ on the part of the defendant or his counsel.’” Ferensic, 501 F.3d at 476 (quoting Lucas, 500 U.S. at 152). In order words, “the exclusion of a defendant’s evidence should be reserved for only those circumstances where ‘a less severe penalty would perpetuate rather than limit the prejudice to the State and the harm to the adversary process.’” Id. (quoting Lucas, 500 U.S. at 152). No. 23-1270 Chandler v. Brown Page 18

Here, much of Chandler’s evidence was purportedly excluded due to violations of discovery rules—for failing to disclose the names of witnesses ahead of trial. These late disclosures were not the result of willful misconduct or other bad-faith reasons; they occurred because Chandler continued to receive discovery until right before his trial. Thus, the trial court’s sanctions were likely inappropriate and disproportionate under the Supreme Court’s case law because of the weighty interests Chandler had at stake.

Although the state court of appeals chastised the trial court for its procedural and evidentiary errors, it rejected Chandler’s constitutional claim on two bases.

First, the court concluded that Chandler had a fair trial because he was represented by counsel. Chandler, 2017 WL 6502801, at *4 n.3. To be sure, denying Chandler counsel would have been unconstitutional, but the mere presence of counsel does not necessarily make a trial constitutionally adequate. Indeed, each of the defendants in Chambers, Washington, and Ferensic had counsel at trial—but the mere presence of competent counsel was insufficient to render their trials fair. Thus, as this court has recognized, the Supreme Court’s case law clearly establishes that Chandler’s claim may not be defeated on the basis that he was represented by counsel at trial.

Second, the state court determined that Chandler had a fair trial because defense counsel “presented defendant’s argument that the victim fabricated the allegations against defendant.” Id. It is true that defense counsel tried to present Chandler’s side of the story through cross- examination—by suggesting that A.C. had a history of false allegations and had a motive to accuse Chandler of abuse. But at every step, the trial court prevented Chandler from producing any evidentiary support for his position, making his defense appear unsubstantiated and perhaps even manufactured. And from the outset, the trial court clearly indicated that that it believed Chandler’s defense to be baseless, telling him: “You’re trying to base a defense here on the fact that, well, she’s lied about all these other things, so that’s my defense for the jury. She’s lying about this. You can argue that if there’s a basis to argue that, but there isn’t at this point.” R. 9- 3, PID 258. No. 23-1270 Chandler v. Brown Page 19

Chambers, Washington, and O’Neal make clear, however, that simply allowing a defendant to raise a defense—even with some evidentiary support—is not always sufficient to ensure a fair trial. In Chambers, for example, eyewitnesses confirmed the defendant’s alibi and testified that McDonald was the shooter. 410 U.S. at 288–89. In Washington, the defendant testified on his own behalf that the accomplice was the actual shooter—testimony that was strengthened by the fact that the accomplice was already convicted of the murder. 388 U.S. at 16. And in O’Neal, where a panel of this court relied on Chambers to grant habeas relief, the defense successfully called several crucial witnesses—one to confirm the defendant’s alibi, a second who witnessed Hickman shoot the victim, and a third who heard the victim’s dying declaration identifying Hickman as the shooter. 933 F.3d at 621–22. Still, in each case, the defendant was denied his right to present a complete defense because the trial court excluded “critical evidence” implicating “constitutional rights directly affecting the ascertainment of guilt.” Chambers, 410 U.S. at 302.

Chandler’s right to present a complete defense was clearly circumscribed even more severely than in Chambers, Washington, and O’Neal. Unlike in those cases, the trial court barred Chandler from calling any witnesses or introducing any evidence on the most critical element of his defense—whether A.C.’s testimony was credible given her history of false allegations. Thus, the state appellate court unreasonably applied the Supreme Court’s governing principles to Chandler’s case and improperly denied his constitutional claim.9 See Williams, 529 U.S. at 413.

The State’s arguments to the contrary are unpersuasive.

a. Prior Precedent

The State first argues that Chambers and Washington do not govern Chandler’s case because they are distinguishable. In Chambers and Washington, the court wrongly excluded evidence showing that another person had committed the charged crime. By contrast, the evidence in Chandler’s case concerned the credibility of the accuser, and there is no precise

9 Because Chandler’s rights were violated by the trial court’s exclusion of Chandler’s witnesses and evidence, we do not decide whether its refusal to delay trial also constituted a constitutional violation. No. 23-1270 Chandler v. Brown Page 20

Supreme Court precedent holding that “the right to present a defense is violated if witnesses are precluded from testifying about an accuser’s past conduct consistent with a common plan or scheme.” Appellee’s Brief at 34. But, as Chandler correctly points out, the State’s argument is better suited to a qualified-immunity analysis, where the “contours of the right must be sufficiently clear” to put state actors on notice. Baynes v. Cleland, 799 F.3d 600, 610 (6th Cir. 2015).

By contrast, AEDPA does not require an “identical factual pattern before a legal rule must be applied.” White v. Woodall, 572 U.S. 415, 427 (2014) (quoting Panetti v. Quarterman, 551 U.S. 930, 953 (2007)). To the contrary, AEDPA assumes that “[c]ertain principles are fundamental enough” to apply to “new factual permutations.”10 Yarborough, 541 U.S. at 666; see also Woodall, 572 U.S. at 427 (“[S]tate courts must reasonably apply the rules ‘squarely established’ by this Court’s holdings to the facts of each case.” (quoting Knowles v. Mirzayance, 556 U.S. 111, 122 (2009))). Thus, the State may not defeat Chandler’s constitutional claim simply because it is premised on new facts. Rather, the State must show that the fundamental principles of the right to present a complete defense that the Supreme Court applied in Chambers and Washington apply differently (or not at all) to Chandler’s case. As we explained above, however, the State is unable to make this showing because the fundamental principle established in Chambers, Washington, and their progeny—that a defendant has the right to present a complete defense—applies with force in Chandler’s case.

b. Duplicative Evidence

The State next argues that Chandler had a fair opportunity to present a defense because the trial court excluded only duplicative or marginal evidence.11 For support, the State cites United States v. Scheffer, 523 U.S. 303 (1998), a case that clarified Chambers’s holding that a constitutional violation occurs when a trial court “significantly undermine[s] the fundamental

10 In Ferensic, for example, the panel granted habeas relief even though the trial court did not exclude evidence of a confession, like in Chambers and Washington, but rather evidence undermining the veracity of the victims’ identifications. Ferensic, 501 F.3d at 475–80. 11 Below, we discuss in detail why the evidence excluded in Chandler’s trial was neither duplicative nor marginal. See infra, Part III(C)(1). That analysis is equally applicable here. No. 23-1270 Chandler v. Brown Page 21

elements of the defendant’s defense.” Id. at 315. In Scheffer, the court-martialed defendant was prohibited from admitting polygraph evidence to bolster his own credibility, but he was allowed to introduce other factual evidence. Id. at 306, 317. The Court concluded that because the “court members heard all the relevant details of the charged offense from the perspective of the accused,” the exclusion of the polygraph did not “implicate any significant interest of the accused.” Id. at 316–17. The State argues that, like in Scheffer, the jury had the opportunity to hear Chandler’s side of the story.

However, Scheffer differs from this case in important ways. Crucially, the Scheffer court excluded only a single piece of evidence, and it had a good reason to do so: polygraph evidence is of questionable reliability. Id. at 306–07, 312 (“Although the degree of reliability of polygraph evidence may depend upon a variety of identifiable factors, there is simply no way to know in a particular case whether a polygraph examiner’s conclusion is accurate, because certain doubts and uncertainties plague even the best polygraph exams.”). Aside from a single piece of potentially unreliable evidence, Scheffer still presented the rest of his defense. Id. at 317. By contrast, the trial court here rushed Chandler to trial, prohibited him from calling any witnesses, and curtailed his ability to acquire and present key evidence. And unlike the polygraph ban in Scheffer, there was no sound rationale to justify the trial court’s actions. See Chandler, 2017 WL 6502801, at *3–4. Thus, the trial court hindered Chandler’s defense far more significantly than in Scheffer.

The question at this stage is not whether Chandler was able to make some of his arguments, but whether the trial court “significantly undermined [the] fundamental elements of [his] defense.” Scheffer, 523 U.S. at 315. Here, there was no physical evidence to support the allegations against Chandler, so A.C.’s testimony was indispensable to his conviction. The trial court, however, “significantly undermined” Chandler’s ability to present evidence casting doubt on her credibility. Id. Chandler’s constitutional claim cannot be defeated simply because the prosecution’s witnesses may have made a few helpful admissions during cross-examination, all of which Chandler was barred from supporting with evidence of his own. Ultimately, the trial court allowed the prosecution to present its side of the story but prevented Chandler from doing the same. No. 23-1270 Chandler v. Brown Page 22

III. Eligibility for Relief

In this circuit we “always” apply Brecht’s “actual prejudice” test in habeas proceedings to assess whether constitutional errors are prejudicial. O’Neal, 933 F.3d at 625. Additionally, we apply the AEDPA standard to a state court’s “harmless beyond a reasonable doubt” assessment of constitutional errors pursuant to Chapman v. California, 386 U.S. 18, 24 (1967). See Brown v. Davenport, 596 U.S. 118, 127 (2022). But we can do so only when the state court actually conducted a Chapman analysis. Here, because the state court concluded that no constitutional error occurred in the first place, it never applied Chapman to Chandler’s case. See Chandler, 2017 WL 6502801, at *4 n.3. Instead, the state court assessed the trial court’s errors under the state’s more forgiving test for non-constitutional errors. Chandler, 2017 WL 6502801, at *4. Under that test, the state court could overturn Chandler’s conviction only if “it affirmatively appear[ed] that it [was] more probable than not that the error was outcome determinative.” Id. (quoting King, 824 N.W.2d at 262). Because we are conducting habeas review of a state-court decision that lacked any analysis under Chapman, we analyze Chandler’s eligibility for relief under the Brecht test alone. See Davenport, 596 U.S. at 138.

Under Brecht, Chandler is entitled to relief if we have “grave doubt[,] not absolute certainty,” Davenport, 596 U.S. at 135, “about whether a trial error of federal law had substantial and injurious effect or influence in determining the jury’s verdict,” Davis v. Ayala, 576 U.S. 257, 267–68 (2015) (quoting O’Neal v. McAninch, 513 U.S. 432, 436 (1995)), see also McAninch, 513 U.S. at 436 (“When a federal judge in a habeas proceeding is in grave doubt about whether a trial error of federal law had ‘substantial and injurious effect or influence in determining the jury’s verdict,’ that error is not harmless. And, the petitioner must win.”). If, “in the judge’s mind, the matter is so evenly balanced that he feels himself in virtual equipoise as to the harmlessness of the error,” then there is “grave doubt” about whether the error affected the jury’s verdict. McAninch, 513 U.S. at 435; see also Davenport, 596 U.S. at 136 (“[W]here AEDPA asks whether every fairminded jurist would agree that an error was prejudicial, Brecht asks only whether a federal habeas court itself harbors grave doubt about the petitioner’s verdict.”). “[P]rosecutors bear the burden of proof” to show that a constitutional error did not have a No. 23-1270 Chandler v. Brown Page 23

substantial and injurious effect on the jury’s verdict. Stoner v. Sowders, 997 F.2d 209, 213 (6th Cir. 1993); see also Jaradat v. Williams, 591 F.3d 863, 869 (6th Cir. 2010) (“Under the Brecht standard, the Government has the burden of showing that the error was harmless.”).12

In applying Brecht, we consider “the whole body of law”—including lower-court cases— to determine whether an error was prejudicial. Davenport, 596 U.S. at 136. And in determining whether the jury’s deliberation was affected by the error, we review the entire record de novo. See Jaradat, 591 F.3d at 869 (“The analysis should result from ‘examination of the proceedings in their entirety.’” (quoting Kotteakos v. United States, 328 U.S. 750, 762 (1946))). The impact of the trial court’s errors depends in part on the strength of the evidence supporting a conviction. Thus, “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.” Towns v. Smith, 395 F.3d 251, 260 (6th Cir. 2005) (quoting Strickland, 466 U.S. at 696).

B. Grave Doubt

Ferensic’s facts are very similar to those here. See Ferensic, 501 F.3d at 470. In Ferensic, the only direct evidence of the defendant’s guilt was the identifications made by the victimized couple. Id. The defense intended to undermine the veracity of the identifications by calling two witnesses, but the trial court barred both from testifying. Id. at 471. The first would have testified that he saw the two culprits on the night of the crime, and that Ferensic resembled neither one; and the second would have testified about the potential unreliability of victim identifications. Id. A panel of this court had grave doubt as to whether the errors affected the jury’s verdict because, without the excluded witnesses, “the jury had no basis beyond defense counsel’s word to suspect the inherent unreliability of the [victims’] identifications.” Id. at 482.

Likewise, the central evidence of Chandler’s guilt came from the testimony of the alleged victim, A.C. As a result, the jury’s verdict likely hinged on the truthfulness of A.C.’s testimony.

12 “[B]ecause ‘[t]he prejudice inquiry is not the same as the sufficiency of the evidence analysis,’” a petitioner cannot be denied relief simply because there is sufficient evidence to support a conviction. Ferensic, 501 F.3d at 474 (quoting Richey v. Mitchell, 395 F.3d 660, 687 (6th Cir. 2005), overruled on other grounds by Bradshaw v. Richey, 546 U.S. 74 (2005)). We are thus prohibited “from simply focusing on the sufficiency of the evidence, especially where it entails ‘stripping the erroneous action from the whole’ and determining the sufficiency of what is left ‘standing alone.’” Id. at 483 (quoting Kotteakos v. United States, 328 U.S. 750, 765 (1946)). No. 23-1270 Chandler v. Brown Page 24

However, the trial court prohibited Chandler from introducing evidence that may have cast reasonable doubt on the reliability of her testimony. In particular, the trial court prevented the jury from considering the following evidence: (1) CPS records detailing at least eight purportedly false allegations that A.C. had made previously, including false allegations of sexual abuse, see supra; (2) testimony from three sets of foster parents—the Hamblins, Nickersons, and Lamberts—who would have discussed A.C.’s prior false allegations; (3) records from another proceeding where Judge Feeney concluded that A.C. had problems with lying and making false accusations; (4) testimony from five foster care workers who had worked with A.C. in the past, all of whom were charged with protecting A.C.’s best interests and still had authored reports detailing her prior false allegations; (5) expert testimony that investigators’ interviewing procedures were faulty, potentially rendering the results unreliable; and (6) expert testimony that it is a “red flag” when a child adds additional details to a story years later, R. 9-10, PID 770. As a result of the trial court’s rulings, the jury had almost “no basis beyond defense counsel’s word to suspect the inherent unreliability” of A.C.’s allegations.13 Ferensic, 501 F.3d at 482.

Weighing the evidence of A.C.’s past allegations against the limited direct evidence supporting Chandler’s conviction, we have grave doubt about the verdict because we think that the excluded evidence could have introduced reasonable doubt into a juror’s mind. See O’Neal, 933 F.3d at 625.

C. The State’s Arguments

The State argues that any error was harmless because (1) the excluded testimony was largely duplicative of testimony given by the prosecution’s witnesses and (2) there was overwhelming evidence to convict Chandler. Neither of these arguments alleviates our grave doubt about whether the trial court’s errors were harmless.

13 The only evidence the jury heard of A.C.’s potential unreliability was the fact that her testimony at trial differed from the accounts she gave investigators in two prior interviews. No. 23-1270 Chandler v. Brown Page 25

1. Duplicative

The State first argues that the trial court’s errors did not affect the jury’s verdict because the excluded evidence was duplicative of other trial testimony. This argument fails for two reasons.

First and foremost, duplicative evidence may strengthen rather than weaken a petitioner’s argument that he was prejudiced by a trial court’s error. For example, the trial court in O’Neal unconstitutionally excluded testimony supporting the defendant’s argument that Hickman shot the victim. 933 F.3d at 627. The Warden in O’Neal argued that the trial court’s errors were not prejudicial because other witnesses gave similar testimony and because O’Neal had the opportunity to cross-examine Hickman. Id. A panel of this court disagreed, explaining that “testimony that mirrors the content of other testimony . . . can still have considerable impact by bolstering the credibility of the other testimony.” Id. In O’Neal, the purportedly duplicative evidence “could have been the straw that broke the camel’s back in establishing a reasonable doubt as to [the defendant’s] guilt.” Id. The panel also rejected the Warden’s argument that cross-examination could cure any potential prejudice, explaining that O’Neal’s ability to impeach witnesses was significantly weaker without the excluded evidence. Id. at 628. Because O’Neal’s logic applies with the same force to Chandler’s case, we reject the State’s assertion that Chandler was not prejudiced simply because the excluded evidence was duplicative or because he had the opportunity to cross-examine A.C.

But even if duplicative evidence were somehow less consequential, the State’s argument would fail because most of the excluded evidence was not duplicative. For example, the State argues that evidence of A.C.’s allegations would have been duplicative at trial because (1) a CPS investigator testified that she was aware that A.C. “had made prior allegations of physical abuse, not of sexual abuse,” and (2) A.C. admitted on cross-examination that she had made two allegations against the Hamblins. R. 9-4, PID 333. However, Chandler was not just trying to show that A.C. made allegations in the past, but also that the allegations were false. Although a CPS investigator testified that she was aware that A.C. had previously made allegations of physical abuse, she did not acknowledge that the allegations were false. Similarly, although No. 23-1270 Chandler v. Brown Page 26

A.C. acknowledged that she made two allegations against the Hamblins, she testified that both were true. Accordingly, the jury heard no evidence that A.C. previously made false allegations.

Had the Hamblins testified, they would have identified specific false allegations that A.C. made. For example, A.C. had alleged that the Hamblins’ dog had attacked her. But when A.C. made this allegation to her counselor, she had no visible bitemarks. Sandy would have testified that, when the counselor confronted A.C., she “giggled, admitted that she lied,” and said, “it’s fun to lie.” R. 9-9, PID 437. The Hamblins would have directly contradicted A.C.’s testimony that she never falsely accused prior foster parents of misconduct. Had Chandler obtained a copy of Judge Feeney’s records in time for trial, the jury would have seen a probate court’s factual findings that A.C. had made false accusations in the past and knew how to “manipulate the system.” R. 9-10, PID 578. And if the foster-care workers were permitted to testify, they would have discussed A.C.’s history of false allegations. Contrary to the State’s argument, this evidence would not have been duplicative, but would have been critical to the jury’s consideration of whether A.C.’s testimony was credible.

The State also argues that the testimony of Kieliszewski, Chandler’s expert, would have mirrored Cottrell’s, the prosecution’s expert. Kieliszewski would have testified that because memories typically degrade rather than strengthen, remembering memories in greater detail over time is a “red flag.” R. 9-9, PID 443. But Cottrell testified during cross-examination that memories generally “degrade” and that people, “particularly young children,” could theoretically come to believe a story that was originally told as a lie. R. 9-5, PID 371. And although Kieliszewski would have discussed RAD, Cottrell answered basic questions about the condition during cross-examination.

However, notwithstanding this overlap, the experts differed in important ways. Although the prosecution’s expert admitted that it is possible for memories to degrade, his testimony was used to advance the argument that memories can become “enriched” as a survivor of assault recounts them repeatedly. R. 9-5, PID 370. The defense expert, by contrast, would have testified that it is a “red flag” for memories to strengthen or change over time because it could indicate that “the alleged victim [has] been coached” or that “they decided to add more to their story because there’s some type of reinforcement they’re getting.” R. 9-10, PID 770. No. 23-1270 Chandler v. Brown Page 27

Kieliszewski would also have testified that this concern is particularly heightened when a victim is re-interviewed several times over the course of years, as occurred with A.C., because “confabulation” could occur with each retelling—where a person may subconsciously fill in the blanks of the memory. R. 9-9, PID 442.

Additionally, Kieliszewski planned to testify about the best practices for conducting forensic interviews with potential victims of child sexual assault, a topic Cottrell did not discuss. The protocols “highly discourage” allowing a support person at an interview because it could affect an interviewee’s answers, but the investigators allowed A.C.’s adoptive mother to sit next to her for the interview. R. 9-9, PID 442. The guidelines also recommend recording forensic interviews, because an investigator’s notes are frequently insufficient to describe an interview comprehensively. The investigator’s report for A.C.’s final interview was just 1.5 pages long, which Kieliszewski said was “quite short” for a ninety-minute interview. Id. Kieliszewski would have also testified that forensic interviews should occur “early after the allegations occur” because they are considered the most reliable. Id. But here, A.C.’s final interview took place years after she made the allegations and differed greatly from her original account.14

As for the RAD-related testimony, the State is correct that the two experts gave overlapping accounts of the condition, which causes children to resist forming bonds with other people. But the testimony about RAD lost its value because the trial court excluded so much other relevant evidence. Chandler attempted to show that A.C. had a motive for making false accusations in the past—she wanted to be removed from each foster family and returned to her biological parents. By explaining the symptoms of RAD, combined with A.C.’s history of making false accusations, Chandler intended to show that A.C.’s accusation was motivated by a desire to avoid adoption and be removed from foster care. Had the trial court allowed the other evidence in, the RAD-related testimony would have played a more significant role.

14 Kieliszewski would have also explained that the guidelines require interviewers to ask questions in a neutral way and to test several different hypotheses with a child interviewee. The purpose of the alternative- hypothesis testing is to avoid bias and to pinpoint the source of potentially unreliable memories—for example, if the allegation is based on something the child saw in a dream or a movie. In the context of child sexual abuse, an interviewer could ask the child about whether the abuse occurred in various environments or whether different people were responsible. By testing alternative hypotheses, investigators can better identify sources of potentially inaccurate memories. Here, A.C.’s interviewers may not have followed the alternate-hypothesis protocol, and Kieliszewski’s testimony would have highlighted concerns about the inconsistency of A.C.’s repeated retellings. No. 23-1270 Chandler v. Brown Page 28

In sum, most of the evidence that the trial court erroneously excluded was not duplicative. Combined with the fact that A.C.’s account changed with each new retelling, we have grave doubt about whether the jury would have convicted Chandler had Chandler been able to present his excluded evidence.

2. Overwhelming Evidence

Finally, the State argues that there was overwhelming evidence of Chandler’s guilt. Three of Chandler’s other alleged victims testified: (1) Chandler’s sister Norma, who alleged that he molested her nearly fifty years prior when they were both children, (2) Z.B.—Stephanie and Lou Jr.’s eight-year-old daughter—who testified that Chandler abused her once or twice each time he saw her, and (3) Harden—Stephanie’s best friend—who testified that Chandler used a “quick swipe” to clean sand “just outside of . . . the genital area” when she was young. R. 9-4, PID 337. Darlene also testified that she once saw Chandler touch A.C.’s leg in a way that “didn’t look right.” R. 9-5, PID 379. Finally, the Chandlers’ family computer contained a written story of child sexual abuse.

However, a juror could reasonably doubt the strength or relevance of this circumstantial evidence after considering the excluded evidence challenging A.C. and others’ credibility. The abuse Norma alleges would have occurred when she and Chandler were children. Harden’s allegation was that Chandler cleaned sand off her, potentially in an inappropriate way, but not that he touched her in the same way as A.C. As for Z.B., her parents were in the middle of a family dispute with Chandler and Darlene. Chandler contends that Lou Jr. was trying to use his parent’s land to begin a marijuana-grow operation and had threatened to kill his mother. Chandler claims that, because of the prosecution, which was triggered by Z.B.’s allegations, he was forced to sell his land. Thus, Chandler argues that Lou Jr. and Stephanie had a motive to urge their daughter to testify against him. Further, one of the witnesses the trial court excluded was Amanda Fraly, who assertedly would have testified that Lou Jr. had pressured her and others to make false allegations against Chandler. And Z.B.’s testimony, like A.C.’s, was inconsistent with the account she gave at an earlier date. No. 23-1270 Chandler v. Brown Page 29

The evidence from the Chandlers’ computer is alarming, but Darlene testified that pornography popped up randomly while she used the computer. Darlene also pointed out that the computer was shared with the couple’s teenage sons, including one who had a history of looking at “real[ly] bad stuff.” R. 9-5, PID 380.

Although the evidence the State cites could support Chandler’s conviction, this is not a sufficiency-of-the-evidence case. As a result, we are “prohibit[ed]” from “focusing on the sufficiency of the evidence, especially where it entails ‘stripping the erroneous action from the whole’ and determining the sufficiency of what is left ‘standing alone.’” Ferensic, 501 F.3d at 483 (quoting Kotteakos, 328 U.S. at 765). Instead, to determine whether the constitutional errors had a “substantial and injurious” effect on Candler’s verdict, we must use a “wider lens” to “ponder[] all that happened” at trial. Id. (quoting Kotteakos, 328 U.S. at 765). The paucity of direct evidence inculpating Chandler on this record, coupled with the considerable excluded evidence casting serious doubt on A.C.’s credibility leaves us with grave doubt about whether Chandler’s verdict was affected by the trial court’s errors.

IV. Conclusion

We REVERSE the district court, conditionally GRANT Chandler’s habeas corpus petition, and REMAND to the district court with instructions to order Chandler’s release from custody unless the State of Michigan grants him a new trial within ninety days.