JOHNSON v. SECRETARY FLORIDA DEPARTMENT OF CORRECTIONS (2025)

This appeal requires us to decide whether admission of a medical report without the author's testimony to support a state prisoner's convictions for kidnapping and sexual battery violated his constitutional right to confront the witnesses against him. See U.S. Const. amend. VI. In 2000, after a man raped 14-year-old C.A., the police took her to a rape treatment center where Dr. Scott Silla examined her. Dr. Silla prepared a report of his findings and about the biological specimens he collected from her. He then gave the specimens and a copy of the report to Detective Steven Signori who sent the specimens to a laboratory for testing. Eight years later, a match in a DNA database linked Reginald Bertram Johnson to the specimens collected from C.A. At trial, the prosecution proved the DNA match through testimony about the collection and testing of the specimens and introduced Dr. Silla's report to bolster the chain of custody for the specimens. Johnson objected that admission of the report violated the Confrontation Clause because Dr. Silla did not himself testify. The jury convicted Johnson. He later filed a petition for a writ of habeas corpus in the district court, which denied the petition. See 28 U.S.C. § 2254. Because Johnson cannot prove actual prejudice, we affirm.

I. BACKGROUND

In 2000, Reginald Bertram Johnson abducted 14-year-old C.A. at gunpoint and raped and robbed her. When C.A. returned home, her mother called the police to report what had happened. The police escorted C.A. to the crime scene so that she could provide more details. And then they brought her to the rape treatment center at Jackson Memorial Hospital.

At the rape treatment center, Dr. Scott Silla conducted a gynecological examination of C.A. and collected biological specimens from her. Nurse Valerie Carter assisted Dr. Silla during this examination. She provided tools and signed a report as a witness. But Nurse Carter did not examine C.A.

Dr. Silla completed a standard report of his findings. The front page of the report included C.A.’s personal information, a description of the assault, the police case number, a drawing of her vaginal tears, and a description of the findings from the pelvic examination. The back page of the report included lists of the tests and treatments performed, an inventory of the specimens collected, and the signatures of Dr. Silla and Nurse Carter.

After the examination, Dr. Silla handed Detective Steven Signori a sealed brown bag with the specimens and a copy of his report. Detective Signori signed the report to confirm that he had taken possession of the specimens. He brought the bag to the Miami-Dade Police Department serology lab.

Sharon Hinz, a forensic analyst at the serology lab, tested the specimens three weeks later. She found semen in the vaginal and cervical swabs. And she discovered a separate DNA profile in addition to C.A.’s in the specimens. But the case went cold for eight years.

In 2008, a routine search of a convicted offender DNA database generated a DNA match. The additional DNA profile from C.A.’s specimens matched Johnson's DNA. Detective Signori met with C.A. and showed her a photograph of Johnson. C.A. stated that she did not know Johnson.

Detective Signori interviewed Johnson and showed him a photograph of C.A. Johnson denied knowing her. Detective Signori then described C.A.’s accusations and showed Johnson pictures of the crime scene. Johnson continued to deny having any sex with C.A. But, after Detective Signori told Johnson that the police had DNA evidence linking him to the crime, Johnson said that “he might have dated her, but he still did not recognize her.”

The police arrested Johnson and collected samples of his saliva. Hinz compared the DNA from these oral swabs to the DNA from the semen collected from C.A. eight years earlier. She confirmed that they matched. The level of accuracy was 1-in-29.8 billion.

The state charged Johnson with two counts of sexual battery with a deadly weapon, one count of armed robbery, one count of armed kidnapping, two counts of armed lewd and lascivious molestation of a child between 12 and 16 years old, and one count of possession of a firearm by a felon. The armed robbery and felony possession charges were later dismissed as barred by the statute of limitations. The rest proceeded to trial in 2011.

At trial, the prosecution's opening statement recounted the details of the rape and ensuing investigation. The prosecutor stated that the “evidence that is really important in this case is the swabs that were taken from her vagina.” And he stated that Johnson's story changed after he was told that his DNA “was found within her vagina.”

Detective Signori testified that Dr. Silla “provided [him] with a copy of the report [Dr. Silla] wrote during the examination” and “a brown paper bag that contained a specimen that [Dr. Silla] obtained during the examination.” He explained that he took the bag to the serology lab for testing. He detailed how the case became cold until 2008 when the police obtained a lead linking Johnson to the crime. He described how Johnson changed his story about possibly knowing C.A. after being told that there was DNA evidence linking him to the crime. And he testified that when he met with Johnson, he noticed that “[h]e had the small bumps on his face that the victim mentioned earlier.” But, on cross-examination, he conceded that C.A.’s physical description of the suspect did not exactly match Johnson and that he did not think that C.A. was “injured” when he met with her.

During Detective Signori's testimony, the prosecution moved to introduce the brown bag containing the specimens into evidence. The defense objected because he could not “testify to the relevance of this item other than [that] the doctor walked out from the examination room and gave him something.” After Detective Signori provided more details, the trial court admitted the rape kit.

Detective Curtis Lueck, who assisted Detective Signori with C.A.’s case in 2008, testified that he was present when Detective Signori interviewed Johnson. And he explained how he collected the oral swabs from Johnson and impounded them in a sealed bag. The trial court admitted that bag into evidence.

Hinz testified that the bag containing the specimens collected from C.A. was sealed when she received it. She explained that she detected semen in the vaginal and cervical swabs. She stated that she found “an additional profile [of DNA] in addition to ‘C.A.’s.’ ” She explained how she later received and tested the oral swabs taken from Johnson in 2008 and produced a report explaining that Johnson's DNA matched the DNA found in the specimens collected from C.A. And she stated that “it would be ․ a one in twenty-nine-point-eight billion chance of finding that profile again.” The trial court admitted Hinz's report into evidence.

Nurse Carter testified that she served as “a witness” for examinations but did not conduct them herself. She stated that she “basically just helped assist [the doctors], giving them the tools that they needed. That's it.” She confirmed that the handwriting on Dr. Silla's report was his, identified the signatures on the report as her own and Dr. Silla's, and explained what various markings on the report meant. She detailed Dr. Silla's “common procedure” for the swabs he collected, which included “collect[ing] the samples,” “put[ting] them in the bag,” “[s]eal[ing] the bag,” and “giv[ing] it to the detective.”

When the prosecution moved to enter Dr. Silla's report into evidence as a business record, the defense objected for lack of foundation and for violations of the Confrontation Clause. The trial court overruled the objection. It later ruled that the report qualified as a business record “because it [was] kept in the normal course of regularly conducted business activity.” And because Nurse Carter had “actual knowledge as to” Dr. Silla's report, she “qualifie[d] as a records custodian.”

Dr. Karen Simmons, the medical director at the rape treatment center when Dr. Silla examined C.A., testified next. Before she testified, the defense objected that her testimony “regarding what is on that report” would violate the Confrontation Clause under Crawford v. Washington, 541 U.S. 36, 124 S.Ct. 1354, 158 L.Ed.2d 177 (2004), and Bullcoming v. New Mexico, 564 U.S. 647, 131 S.Ct. 2705, 180 L.Ed.2d 610 (2011), because “[s]he was not present for any of the examination.” The trial court invited the defense to “object whenever you think it is appropriate.” Dr. Simmons explained that she trained and worked with Dr. Silla and Nurse Carter. She verified that Dr. Silla's initials were on the brown bag and swabs. She explained that the purpose of examinations at the rape treatment center is “diagnosis and treatment of that patient.” Over the defense's objection, she interpreted Dr. Silla's report and explained the significance of “fresh tears” to the vagina as “evidence of blunt penetrating trauma.”

C.A. testified last. She recounted the details of the rape, including that she was 14 years old at the time and saw Johnson holding “a small, black handgun.” She testified that he forced her to have oral and vaginal sex with her and stole her money. And she remembered how Johnson had finally let her go after she begged for her life: “Please don't kill me. Please just let me go. I promise I won't tell nobody.”

C.A. also described her examination at the rape treatment center. She recalled that the doctor took specimens from her and that a female medical professional was also present. She explained that she felt pain when using the bathroom and observed blood in her vaginal area. On cross-examination, she admitted that she was not sure how many specimens were collected from her, nor did she recall Dr. Silla taking any blood from her.

After C.A.’s testimony, the defense moved for a judgment of acquittal. The trial court granted the motion to the extent that the counts relied on C.A. being threatened with a knife. It also granted the motion as to one of the lewd and lascivious molestation counts because it found that it was identical to the other count of the same name. But it denied the motion as to the remaining counts. The defense presented no witnesses, and Johnson did not testify.

The prosecutor argued in closing that Detective Signori “g[o]t the one thing that he needed, which was the DNA” and that this “piece of evidence is huge.” The prosecutor recounted how multiple witnesses testified about the collection of the DNA sample from C.A. and how it matched Johnson's DNA. She described the findings contained in Dr. Silla's report and asked the jury “to look at it with a fine tooth comb.” She stated that the “[f]resh tears on a fourteen year old's vaginal area,” as described by the report, were “one of the most important things here.” She recounted that C.A. also testified about how she felt these injuries. And she argued that “we have the best piece of evidence possible, his DNA”; “[t]he value of the DNA, the weight that DNA carries is all that really matters”; and “DNA does not lie.”

The jury convicted Johnson of two counts of sexual battery with a deadly weapon, one count of kidnapping, and one count of lewd and lascivious molestation. The court sentenced Johnson to life sentences for the sexual battery and kidnapping counts and 30 years of imprisonment for the lewd and lascivious molestation count. It directed that the sentences on each count would run consecutively.

Johnson appealed his convictions and sentences. Johnson v. State, 117 So. 3d 1238, 1239 (Fla. Dist. Ct. App. 2013). He challenged the introduction of the rape kit and Dr. Silla's report. Id. The appellate court held that the rape kit and report were properly authenticated as business records, see FLA. STAT. § 90.803(6) (2014), and that admission of the report did not violate the Confrontation Clause because Johnson had the opportunity to cross-examine “the person who actually performed the DNA test, Sharon Hinz,” and “Nurse Carter, who was present and assisted Dr. Silla in his examination and his collection of the evidence, [and] also signed the report.” Johnson, 117 So. 3d at 1239, 1244–45. Finally, the court ordered vacatur of Johnson's conviction for lewd and lascivious molestation for violating the Double Jeopardy Clause. Id. at 1245.

The Supreme Court of Florida denied Johnson's petition for discretionary review. Johnson v. State, 147 So. 3d 524 (Fla. 2014) (table). The Supreme Court of the United States denied Johnson's petition for a writ of certiorari. Johnson v. Florida, 574 U.S. 896, 135 S.Ct. 239, 190 L.Ed.2d 180 (2014) (mem.). Johnson then exhausted his state post-conviction remedies.

In 2019, Johnson filed pro se an amended petition for a writ of habeas corpus in the district court. See 28 U.S.C. § 2254. He alleged seven claims for relief, including that admission of Dr. Silla's report violated the Confrontation Clause. The magistrate judge recommended denial of the petition. She concluded that “there was no Confrontation Clause violation because Hinz's testimony that she examined and tested the evidence, matching the DNA profile from the samples taken from C.A. to [Johnson]’s DNA profile, was subjected to extensive cross-examination by the defense regarding her testing and findings.” Despite Johnson's objection, the district court adopted the magistrate judge's report and recommendation and denied Johnson's petition.

We granted a certificate of appealability on one issue: “Whether the district court erred by denying Johnson's [c]laim ․ that the trial court violated his confrontation rights by admitting into evidence a forensic analysis report, containing testimonial hearsay, when its author was not called to testify at trial and, thus, could not be cross-examined.”

II. STANDARD OF REVIEW

In an appeal of a denial of a petition for a writ of habeas corpus, we review whether an alleged constitutional error is harmless de novo. Mason v. Allen, 605 F.3d 1114, 1123 (11th Cir. 2010).

III. DISCUSSION

State prisoners “are not entitled to habeas relief based on trial error unless they can establish that it resulted in actual prejudice.” Brecht v. Abrahamson, 507 U.S. 619, 637, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993) (citation and internal quotation marks omitted). “Under this test, relief is proper only if the federal court has grave doubt 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, 135 S.Ct. 2187, 192 L.Ed.2d 323 (2015) (citation and internal quotation marks omitted). This standard is “extremely demanding.” Al-Amin v. Warden, Ga. Dep't of Corr., 932 F.3d 1291, 1303 (11th Cir. 2019).

As a preliminary matter, we address an apparent tension in our precedents about the order of operations when we deny a habeas petition. On the one hand, several of our decisions suggest that we cannot consider actual prejudice until after we have concluded that there was an underlying error. In Williams v. Singletary, we stated that because “[a] harmless error inquiry necessarily entails a two-step process,” “[a] court must first find an error before it can determine whether that error is harmless.” 114 F.3d 177, 180 (11th Cir. 1997). And, in Al-Amin v. Warden, Georgia Department of Corrections, we similarly announced that “we must first find an error before we can determine whether that error is harmless” in an appeal from the denial of a habeas petition. 932 F.3d at 1302.

On the other hand, some of our decisions have resolved habeas appeals on lack of prejudice without first holding that there was an underlying error. In Hodges v. Attorney General, State of Florida, we explained that “[w]e need not decide whether the admission ․ of the testimony ․ violated the Confrontation Clause” because, “[e]ven assuming that it did, any violation was harmless.” 506 F.3d 1337, 1342–43 (11th Cir. 2007). And, in Trepal v. Secretary, Florida Department of Corrections, we “assume[d], but d[id] not decide, ․ that [the petitioner] ha[d] satisfied th[e] threshold requirements of § 2254(d)” because we could deny his petition “under the standard set forth in Brecht.” 684 F.3d 1088, 1113–14, 1113 n.30, 1117 (11th Cir. 2012).

Although language from these precedents appears to conflict, a closer look establishes that any suggestion that we must find an underlying error before considering actual prejudice is dicta. And, to the extent that any conflict between holdings exists, the latter view that we can skip directly to the actual-prejudice inquiry must prevail in the light of our prior-panel-precedent rule and Supreme Court precedent in adjudicating habeas appeals.

To start, the statements in Williams and Al-Amin about needing to find an error before considering actual prejudice are nonbinding dicta. Dicta are “those portions of an opinion that are not necessary to deciding the case then before us.” United States v. Caraballo-Martinez, 866 F.3d 1233, 1244 (11th Cir. 2017) (citation and internal quotation marks omitted); see also Bryan A. Garner et al., The Law of Judicial Precedent § 4, at 46 (2016) (“Generally, a dictum is a statement in a judicial opinion that is unnecessary to the case's resolution.”). Because our prior-panel-precedent rule “applies only to holdings, not dicta,” we are not bound by these types of statements. United States v. Birge, 830 F.3d 1229, 1232 (11th Cir. 2016).

Our declaration in Williams that “[a] court must first find an error before it can determine whether that error is harmless” was dictum. 114 F.3d at 180. There, we affirmed the denial of the petitioner's habeas petition because he failed to satisfy the four-element test for establishing that a refusal to sever a codefendant rendered his trial fundamentally unfair. Id. at 179–81. The petitioner argued that Brecht supplanted the standard for the third element of the substantive standard because both involved harmless error. Id. at 179–80. We rejected this argument because “[n]othing in Brecht suggests its harmless error standard should apply at the first step of the inquiry and change what it takes to establish an error in the first place.” Id. at 180. In reaching this holding, we explained that because “[a] harmless error inquiry necessarily entails a two-step process,” it would not make sense to blend the substantive standard at step one with Brecht’s actual-prejudice inquiry at step two. Id. We also stated that “[a] court must first find an error before it can determine whether that error is harmless.” Id. But this statement was dictum because it was “not necessary to deciding the case then before us.” Caraballo-Martinez, 866 F.3d at 1244 (citation and internal quotation marks omitted). We would have reached the same conclusion without it because what mattered was that “[a] harmless error inquiry necessarily entails a two-step process”—not the order that courts apply that process. Williams, 114 F.3d at 180.

Our statement in Al-Amin that “we must first find an error before we can determine whether that error is harmless” was also dictum. 932 F.3d at 1302. There, we affirmed the denial of the petitioner's Confrontation Clause claim in his habeas petition because there was no underlying error. Id. at 1302–03. This analysis was limited to Brecht’s framework and did not consider section 2254(d). See id. So this statement has no bearing on the order in which we address section 2254(d)’s statutory requirements versus Brecht’s actual-prejudice inquiry. And, to the extent that the statement could be read to require a finding of error in the Brecht framework before moving to its actual-prejudice inquiry, it is still dictum because announcing this order was merely a “prefatory statement” that was “not necessary to our holding.” Ga. Ass'n of Latino Elected Offs., Inc. v. Gwinnett Cnty. Bd. of Registration & Elections, 36 F.4th 1100, 1120 (11th Cir. 2022). To analyze whether an error was harmless, we must know what the error was. But “find[ing]” an error does not mean that we must determine that an error was committed. Al-Amin, 932 F.3d at 1302. Instead, we can identify an error by assuming that the error the petitioner identifies was an error.

Next, even if we treated these statements in Williams and Al-Amin as holdings, they are not binding because earlier precedents held that we may deny a habeas petition for lack of actual prejudice without deciding the existence of an underlying error. “When we have conflicting case law, we follow our oldest precedent.” United States v. Madden, 733 F.3d 1314, 1319 (11th Cir. 2013). Williams appears to be our oldest decision that comments on the order of Brecht’s actual-prejudice inquiry in this manner. But our statement in Williams that “[a] court must first find an error before it can determine whether that error is harmless” could not have established a binding rule prohibiting us from resolving a habeas appeal on actual-prejudice grounds alone because earlier precedents allowed that practice. 114 F.3d at 180. In Williams, we expressly cited as support our decisions in Horsley v. Alabama, 45 F.3d 1486, 1492 (11th Cir. 1995), where we “assum[ed] [constitutional] error but f[ound] error to be harmless under Brecht,” and Bolender v. Singletary, 16 F.3d 1547, 1566–67 (11th Cir. 1994), where we found no constitutional violation “but h[eld] in the alternative that any such violation would be harmless under Brecht.” Williams, 114 F.3d at 180. Because these precedents predated Williams, it could not have overturned existing practice. That Williams relied on precedents that are arguably inconsistent with its general statement further confirms that its language about the order for evaluating actual prejudice was an “aside-like statement[ ]” that has no binding effect. United States v. Files, 63 F.4th 920, 929 (11th Cir. 2023).

Finally, the statements in Williams and Al-Amin are inconsistent with intervening Supreme Court precedent. In Brown v. Davenport, the Supreme Court reiterated that “a federal court must deny relief to a state habeas petitioner who fails to satisfy either this Court's equitable precedents [like Brecht] or AEDPA.” 596 U.S. 118, 142 S. Ct. 1510, 1524, 212 L.Ed.2d 463 (2022) (second emphasis added). It explained that these “inquiries are entirely different in kind.” Id. at 1525 (citation and internal quotation marks omitted); see also Lukehart v. Sec'y, Fla. Dep't of Corr., 50 F.4th 32, 42 (11th Cir. 2022) (agreeing that “[t]his harmless-error inquiry is separate from the AEDPA analysis”). And it confirmed that, “if a federal court determines that a habeas petition fails because of Brecht, there is no need to prolong the matter by ‘formally applying’ AEDPA as well.” Davenport, 142 S. Ct. at 1527 (alterations adopted) (quoting Fry v. Pliler, 551 U.S. 112, 120, 127 S.Ct. 2321, 168 L.Ed.2d 16 (2007)). Similarly, in Davis v. Ayala, the Court resolved a habeas appeal by “[a]ssuming without deciding that a federal constitutional error occurred” because “the error was harmless under Brecht.” 576 U.S. at 260, 135 S.Ct. 2187.

Because we can resolve this appeal based on lack of actual prejudice under Brecht, we assume—but do not decide—that admission of Dr. Silla's report violated Johnson's constitutional right to confront the witnesses against him. See U.S. Const. amend. VI. The actual-prejudice inquiry turns on whether we have “grave doubt” that the jury would have convicted Johnson without admission of Dr. Silla's report. Ayala, 576 U.S. at 268, 135 S.Ct. 2187 (citation and internal quotation marks omitted). “Although harmless error review is necessarily fact-specific and must be performed on a case-by-case basis, the erroneous admission of evidence is likely to be harmless under the Brecht standard where there is significant corroborating evidence or where other evidence of guilt is overwhelming.” Mansfield v. Sec'y, Dep't of Corr., 679 F.3d 1301, 1313 (11th Cir. 2012) (citations omitted). Johnson contends that the introduction of the report was not harmless because Dr. Silla “was undoubtedly the state's most important witness” and there was a lack of corroborating evidence that the specimens that later proved the DNA match with Johnson came from C.A. The Secretary responds that any error was harmless because the DNA match was thoroughly corroborated regardless of the report based on the testimony of C.A., Nurse Carter, Detective Signori, and Hinz about the collection, transport, and testing of the specimens collected at the rape treatment center. The Secretary has the better argument.

Johnson fails to establish actual prejudice because other chain-of-custody testimony independently proved that the specimens collected from C.A. contained Johnson's DNA. Even if the trial court had excluded the report, there was still sufficient testimony for the jury to conclude beyond a reasonable doubt that Johnson was the perpetrator due to the DNA match. C.A. testified about how the doctor collected specimens from her. Detective Signori testified that Dr. Silla gave him the specimens—in a sealed bag—after the examination, which he then impounded at the serology lab. Hinz testified that she took those impounded samples, which “ha[d]n't been opened by anyone else,” and ran tests on them that later matched DNA collected from Johnson. And Nurse Carter testified that she signed a report that described C.A.’s examination and the collection of specimens from her. This chain-of-custody evidence is “significant corroborating evidence” of the report's statement that Dr. Silla collected the specimens from C.A. Id.

Because “a rational jury could consider the DNA evidence to be powerful evidence of guilt,” this DNA evidence linking Johnson to the rape supported the jury's verdict regardless of the report. McDaniel v. Brown, 558 U.S. 120, 132, 130 S.Ct. 665, 175 L.Ed.2d 582 (2010). As the Secretary explains, “even excluding the report, Johnson had no plausible explanation for how his semen happened to be in police custody. The only rational conclusion for a jury to reach was that his semen was tied to this rape.” So, even if there were some inconsistencies between C.A.’s recollection of the number and type of specimens taken from her by Dr. Silla and other evidence presented at trial, Johnson's failure to rebut the DNA match mitigated any risk of prejudice.

Other evidence also corroborated the prosecution's case. C.A.’s testimony about the rape gave the jury a legitimate basis to conclude that the prosecution proved all elements of the crimes beyond a reasonable doubt. That Johnson changed his story when confronted with the claim that his DNA was found inside C.A. supported the jury's verdict. And Detective Signori's testimony that the small bumps on Johnson's face matched C.A.’s description of the rapist provided additional corroboration. Taken together with the DNA match, the evidence against Johnson does not leave us with “grave doubt about whether” admission of the report “had substantial and injurious effect or influence in determining the jury's verdict.” Ayala, 576 U.S. at 268, 135 S.Ct. 2187 (citation and internal quotation marks omitted).

Johnson's contentions to the contrary are unpersuasive. Johnson argues that, “[i]n h[er] closing argument, the prosecutor emphasized the importance of Dr. Silla's report to the state's case against Johnson.” But the record shows that the prosecution's case primarily hinged on the DNA match, not the report itself. In her closing argument, for example, the prosecutor emphasized that “we have the best piece of evidence possible, his DNA,” and that “DNA does not lie.” That she also asked the jury “to look at [Dr. Silla's report] with a fine tooth comb” does not amount to actual prejudice because she made clear that “[t]he value of the DNA, the weight that DNA carries is all that really matters.” Johnson also asks, “If the state didn't need Dr. Silla's report to convict Johnson, why did the state call Dr. Simmons for the sole purpose of parroting the content of Dr. Silla's report to the jury[?]” But that the prosecution provided additional evidence in one form does not mean that the other evidence would not have been enough on its own. In this appeal, the DNA match and other evidence presented at trial meant that “other evidence of guilt” was “overwhelming.” Mansfield, 679 F.3d at 1313. Because Johnson cannot prove actual prejudice, we affirm the denial of his petition.

IV. CONCLUSION

We AFFIRM the denial of Johnson's petition.

In discussing the Brecht v. Abrahamson, 507 U.S. 619, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993) harmless-error standard for habeas cases, we explained in Williams v. Singletary, 114 F.3d 177 (11th Cir. 1997) that “[a] court must first find an error before it can determine whether that error is harmless.” Id. at 180. The majority opinion concludes that this explanation from Williams “was dictum.”¹ I'm not so sure.

In Williams, the habeas petitioner sought “relief for a trial court's refusal to sever a codefendant.” 114 F.3d at 179. Under our precedent for that kind of severance claim, the petitioner had to “show that the refusal rendered the trial fundamentally unfair.” Id. A decade earlier, in Smith v. Kelso, 863 F.2d 1564 (11th Cir. 1989), we “established a four-step test for determining whether a defendant's nonsevered trial was fundamentally unfair.” Williams, 114 F.3d at 179. The third step asked whether “the conflict subject[ed] the [petitioner] to compelling prejudice?” Id. (quoting Smith, 863 F.2d at 1568). On appeal from the denial of his habeas petition, the Williams petitioner argued that the district court “applied the wrong legal standard in the third step of the Smith analysis” because “the Supreme Court effectively overruled” our test “and instituted a new rule of decision in Brecht.” Id. (citation omitted). The petitioner “contend[ed] that the Brecht harmless[-]error standard supplanted the ․ compelling prejudice inquiry.” Id. at 180.

The Williams court “reject[ed]” the petitioner's argument that the Brecht harmless-error standard supplanted the compelling prejudice inquiry and gave one reason for the rejection—“[a] court must first find an error before it can determine whether that error is harmless.” Id. “Nothing in Brecht,” we explained, “suggests its harmless[-]error standard should apply at the first step of the inquiry and change what it takes to establish an error in the first place.” Id. Thus, Williams continued, “[a] habeas petitioner must satisfy th[e] four-step test before moving on to the Brecht harmless[-]error standard.” Id. We explained that we would “not stretch Brecht beyond its holding, which is that the [harmless-error] standard applies in determining when an already established constitutional violation nonetheless amounts to harmless error.” Id.

Our reading of Brecht in Williams as applying the harmless-error standard when a constitutional violation had been “already established” was necessary to our rejection of the petitioner's argument that the Brecht harmless-error standard supplanted our compelling prejudice inquiry. See id. That a court must first find error before it can apply the Brecht harmless-error standard was the one reason why we reached the result that we reached in Williams and, thus, was a necessary part of our holding. See United States v. Caraballo-Martinez, 866 F.3d 1233, 1244 (11th Cir. 2017) (“The holding of a case comprises both the result of the case and those portions of the opinion necessary to that result.” (quotation omitted)); Bryan A. Garner et al., The Law of Judicial Precedent § 4, at 44 (Thomas Reuters 2016) (“A holding consists of the court's determination of a matter of law pivotal to its decision.” (quotation and footnote omitted)).

Yet, despite our different views on Williams, I get to the same place as the majority opinion because Williams told us what it means for a court to “first find an error before it can determine whether that error was harmless.” 114 F.3d at 180. The Williams court explained that “[w]e have consistently applied the Brecht harmless[-]error standard only after determining that there was an error,” and then gave nine examples of how that works in practice. Id. One example was from Horsley v. Alabama, 45 F.3d 1486 (11th Cir. 1995). There, we “[a]ssum[ed]” error but “nevertheless affirm[ed] the district court's denial of relief under the doctrine of harmless error.” Id. at 1492. The Williams court cited and described this assumed-error approach as “consistent[ ]” with applying the Brecht harmless-error standard only after determining that there was an error. See Williams, 114 F.3d at 180. That is, under Williams, one way to comply with the directive that “[a] court must first find an error before it can determine whether that error is harmless” is to assume error, as we did in Horsley, and then decide whether any error was harmless under Brecht. See Williams, 114 F.3d at 180.

Because assuming error complies with Williams’s directive that courts “must first find an error before it can determine whether that error is harmless,” I don't see any tension in our precedents. See id. We did the same thing in Hodges v. Attorney General, State of Florida, 506 F.3d 1337 (11th Cir. 2007). There, we “assum[ed]” a Confrontation Clause violation at the habeas petitioner's trial but concluded that “any violation was harmless.” Id. at 1342–43. And we did the same thing in Trepal v. Secretary, Florida Department of Corrections, 684 F.3d 1088 (11th Cir. 2012). In Trepal, we “assum[ed]” a Giglio v. United States, 405 U.S. 150, 92 S.Ct. 763, 31 L.Ed.2d 104 (1972) violation but held that “the Giglio error [was] harmless under the standard set forth in Brecht.” Trepal, 684 F.3d at 1113–14 & n.30. Our assumed-error approach to Brecht in Hodges and Trepal was the same one we gave as an example of a “consistent[ ]” application of Brecht’s harmless-error standard in Williams. See 114 F.3d at 180.

I also do not see how Williams is inconsistent with intervening Supreme Court precedent.² In Davis v. Ayala, 576 U.S. 257, 135 S.Ct. 2187, 192 L.Ed.2d 323 (2015), the Supreme Court took the same assumed-error approach to Brecht that we blessed in Williams. The Davis Court “[a]ssum[ed] without deciding that a federal constitutional error occurred” but concluded that “the error was harmless under Brecht.” Id. at 260, 135 S.Ct. 2187 (citation omitted).

The majority opinion is also consistent with the assumed-error approach to Brecht that we said in Williams was consistent with applying the harmless-error standard “only after determining that there was an error.” See 114 F.3d at 180. The majority opinion “assume[s],” but does not decide, “that admission of Dr. Silla's report violated Johnson's constitutional right to confront the witnesses against him,” but then goes on to explain why any error was harmless under Brecht. Because I think this approach is consistent with Williams, our precedent, and Supreme Court precedent, and because I think the majority opinion is right that under the circumstances of this case any Confrontation Clause error did not have a substantial and injurious effect or influence in determining the jury's verdict, I join the judgment affirming the denial of Reginald Johnson's habeas petition.

FOOTNOTES

1.   I agree with the majority opinion that our statement in Al-Amin v. Warden, Georgia Department of Corrections, 932 F.3d 1291 (11th Cir. 2019) that “we must first find an error before we can determine whether that error is harmless” was dicta in that case because it was not necessary to the holding. See id. at 1302.

2.   I do not read the statement in Brown v. Davenport, 596 U.S. 118, 142 S.Ct. 1510, 212 L.Ed.2d 463 (2022) that “a federal court must deny relief to a state habeas petitioner who fails to satisfy either th[e Supreme Court's] equitable precedents or AEDPA” as inconsistent with Williams’s holding about the order of operation under Brecht. Id. at 134, 142 S. Ct. 1510, 1524. Of course, as Brown explains, where a state court decides a federal constitutional claim on the merits, the petitioner must satisfy the AEDPA standard and Brecht to be entitled to habeas relief in federal court because the AEDPA and Brecht “inquiries are entirely different in kind.” Id. at 135, 142 S. Ct. 1510 (citation and quotation omitted). But Brown does not say anything about the proper order of operation once the federal court chooses to decide the petition under the Brecht harmless-error standard. Williams does. See 114 F.3d at 180.

William Pryor, Chief Judge: