GREGG COSTA, Circuit Judge:
A Mississippi jury convicted Lisa Jo Chamberlin of two counts of capital murder and sentenced her to death. The district court granted Chamberlin’s petition for writ of habeas corpus on the ground that the state court erred in finding that there was no racial exclusion of jurors at her trial. We affirm.
I.
Even by the standards of capital cases, the double murder in this case was gruesome. It occurred in Hattiesburg, Mississippi, where Chamberlin and her then boyfriend, Roger Gillett, had recently moved in with Gillett’s cousin, Vernon Hulett, and Hulett’s girlfriend, Linda Heintzelman. After an argument, Hulett and Heintzelman told Chamberlin and Gillett to move out. Unwilling to leave, Gillett began beating Hulett and Heintzelman and demanded that Hulett tell him the combination to a safe in Hulett’s bedroom. Although Hulett eventually disclosed a combination, no one was able to open the safe. In an escalating rage, Gillett continued to physically assault Hulett, and he and Chamberlin physically and sexually assaulted Heintzelman.
Eventually, Chamberlin told Gillett they should murder Hulett and Heintzelman and flee. Gillett struck Hulett in the head with a hammer and slashed his throat. Chamberlin attempted to strangle Heint-zelman but was not strong enough to suffocate her, so Gillett stabbed Heintzelman. Chamberlin and Gillett left the home to dispose of the knife and hammer. When they returned, Heintzelman was lying on the floor, still breathing. After leaving her there for most of the day, the couple finally decided to suffocate her. They bound her hands with duct tape and put plastic bags over her head until she stopped breathing.
The couple put both bodies in Hulett’s freezer, and, taking with them the freezer and all the evidence they could collect, they left Mississippi. They ended up in Kansas, where Gillett had family. Almost immediately, Kansas authorities took them into custody on drug charges and obtained a search warrant for a farm where authorities suspected the couple was manufacturing crystal meth. During the search, sher[660]*660iffs deputies found the bodies in the freezer.
Back in Mississippi, Gillett and Cham-berlin were tried separately. Both trials resulted in death sentences, though Gil-lett’s sentence was vacated on state post-conviction review.
II.
Even for the most horrific of crimes with the most culpable of defendants, there are certain trial errors that are deemed structural and require automatic reversal. Sullivan v. Louisiana, 508 U.S. 275, 281, 113 S.Ct. 2078, 124 L.Ed.2d 182 (1993). The ground on which the district court granted Chamberlin relief, exclusion of jurors on racial grounds, is one example. “Discrimination in jury selection ... causes harms to the litigants, the community, and the individual jurors who are wrongfully excluded from participation in the judicial process.” J.E.B. v. Alabama ex rel. T.B., 511 U.S. 127, 140, 114 S.Ct. 1419, 128 L.Ed.2d 89 (1994). Going all the way back to one of its first cases finding a violation of the Equal Protection Clause (Strauder v. West Virginia, 100 U.S. 303, 312, 10 Otto 303, 25 L.Ed. 664 (1880)), the Supreme Court thus “has followed an automatic reversal rule once a violation of equal protection in the selection of jurors has been proven.” Winston v. Boatwright, 649 F.3d 618, 627 (7th Cir. 2011); see also Scott v. Hubert, 610 Fed.Appx. 433, 434 (5th Cir. 2015) (“[Discrimination on the basis of race in the selection of ... jurors is a form of structural error that voids a conviction.”).1 And because such error “casts doubt on the integrity of the judicial process and places the fairness of a criminal proceeding in doubt,” a defendant may challenge the exclusion of jurors of a different race. Powers v. Ohio, 499 U.S. 400, 406-11, 111 S.Ct. 1364, 113 L.Ed.2d 411 (1991) (internal citations omitted).
Chamberlin, a white defendant, invokes that rule to challenge the exclusion of black jurors. After the trial judge narrowed an initial pool of several hundred prospective jurors to 42 qualified jurors, 31% of whom were black, both sides exercised peremptory strikes. The prosecutor first went through the list of qualified prospective jurors in order, striking and accepting jurors as he went, until the State proffered a prospective jury of twelve. The defense then had an opportunity to strike or accept the proffered jurors.
The prosecutor struck two of the first three black jurors and accepted eleven of the first twelve white jurors, proffering an initial proposed jury of eleven white jurors and one black juror. After defense counsel struck several of those jurors, the State continued in the same manner, striking the next five black jurors (including Thomas Sturgis and David Minor who will become important), before accepting two black jurors. Even this low number is more than the State planned to accept. The prosecutor believed the second black juror had been struck for cause prior to the peremptory phase.
Ultimately, the prosecutor used eight of his thirteen strikes,2 or 62%, against black [661]*661jurors. Ten white jurors and two black jurors sat on the jury; both alternates were white.
Chamberlin objected to the strikes, arguing they constituted a prima facie case of discrimination under Batson v. Kentucky, 476 U.S. 79, 106 S.Ct. 1712, 90 L.Ed.2d 69 (1986), which established a framework for determining if peremptory strikes are racially motivated. Applying Batson, the court asked the prosecutor if he had race-neutral reasons for the strikes. For the jurors relevant to this appeal, Sturgis and Minor, the prosecutor pointed to their answers to three questions on written questionnaires the jurors had completed before trial. The prosecutor claimed he struck them because of the answers they checked in response to questions 30, 34, and 35, in which both stated: (1) they were “not sure” if they were emotionally capable of announcing a verdict of death; (2) they were “not sure,” because it was a capital case, if they would hold the State to a higher burden of proof than the law requires; and (3) “yes,” because the defendant faced the death penalty, they would want to be one hundred percent certain before finding the defendant guilty.3
The trial court accepted these race-neutral reasons. Defense counsel responded by noting that Sturgis generally favored the death penalty and that Minor had no opinion on the death penalty, and, like other jurors the prosecutor had accepted, Minor had a relative in law enforcement. Based on “the totality of their questionnaire[s],” defense counsel argued, “it appears that they could be absolutely open- and fair-minded jurors on the question of the death penalty.” Defense counsel also pointed out that the State had not sought to question Sturgis or Minor individually to follow up on their questionnaires. Without commenting on the defense’s arguments, the trial court rejected the Batson claim.
The court did not conduct a “comparative juror analysis”: an analysis of whether reasons given by the prosecutor for striking black jurors apply equally to white jurors the prosecutor accepted. See Reed v. Quarterman, 555 F.3d 364, 369 (5th Cir. 2009). Chamberlin did not point out, and the court did not consider, that a white juror the prosecutor had accepted, Brannon Cooper, gave the same answers as Sturgis and Minor to questions 30, 34, and 35. Like Sturgis and Minor, Cooper was “not sure” if he was “emotionally capable of standing up in court and announcing [a] verdict as to the defendant being put to death.” Like them, he was “not sure” if he would “hold the state to a greater burden of proof than the law requires because this case is one in which the death penalty may be imposed.” And, “because this case involves the death penalty,” he would “want to be 100% certain” of guilt before returning a guilty verdict. Despite the three men giving the same answers, the prosecutor accepted Cooper [662]*662yet struck Sturgis and Minor.4
On direct appeal, Chamberlin claimed she was entitled to relief on six grounds, including Batson. Chamberlin v. State, 989 So.2d 320 (Miss. 2008). Chamberlin again did not compare Sturgis and Minor to Cooper. Without conducting a comparative juror analysis, the Supreme Court of Mississippi denied relief, finding with regards to Sturgis and Minor that “the defense did not meet its burden to show that the facts and circumstances give rise to the inference that the prosecutor exercised the peremptory challenges with a discriminatory purpose.” Id. at 339.5
In her federal petition, Chamberlin asserted she was entitled to relief on thirteen grounds, including that the state court clearly erred when it found there was no Batson violation. Chamberlin v. Fisher, 2015 WL 1485901, at *12 n.3 (S.D. Miss. Mar. 31, 2015). The district court agreed: a Batson violation had occurred which warranted vacating Chamberlin’s conviction and sentence. Id. at *21-23. The State appeals.
III.
The district court granted the writ in a proceeding governed by the Antiterrorism and Effective Death Penalty Act (AED-PA). It found that Chamberlin’s Batson claim warranted federal relief under either of the two grounds on which a federal court can grant a writ based on a claim that was decided in state court. Those are when the state court adjudication:
(1) resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States; or
(2) resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.
28 U.S.C. § 2254(d); see Wiggins v. Smith, 539 U.S. 510, 520, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003).
The district court held that the legal error subsection (d)(1) describes existed because the Supreme Court of Mississippi did not conduct the comparative juror analysis used in Miller-El v. Dretke, 545 U.S. 231, 125 S.Ct. 2317, 162 L.Ed.2d 196 (2005) (Miller-El II). It also held that the factual error subsection (d)(2) describes existed because, after using comparative juror analysis, it concluded that the state court’s finding of race-neutral strikes for Sturgis and Minor was unreasonable.
We need not reach the section 2254(d)(1) question whether the state court contravened Miller-El II in failing to conduct a [663]*663comparative juror analysis. Compare McDaniels v. Kirkland, 813 F.3d 770, 777 (9th Cir. 2015) (en banc) (not deciding whether Miller-El II requires state courts to conduct comparative juror analysis when reviewing Batson claims), with id. at 782 (Ikuta, J., concurring) (arguing that “Miller-El II could not and did not establish any such rule” requiring state courts to conduct such analysis, though recognizing that comparative juror analysis may be used in conducting the section 2254(d)(2) inquiry). That is because we affirm the judgment on the ground that the Mississippi court’s decision was based on an unreasonable determination of the facts under section 2254(d)(2). See Miller-El II, 545 U.S. at 240, 266, 125 S.Ct. 2317 (granting relief under section 2254(d)(2)). In conducting that factual review, both the Supreme Court and this court have used the comparative juror analysis even when state courts rejecting the Batson claim never did. See id. at 241, 241 n.2, 125 S.Ct. 2317 (conducting comparative analysis on habeas review despite no such analysis being presented to state courts); Reed, 555 F.3d at 372-73 (same); Woodward v. Epps, 580 F.3d 318, 338 (5th Cir. 2009) (same).
A court may grant relief for the factual error section 2254(d)(2) captures when it concludes that the state court’s “decision was unreasonable or that the factual premise was incorrect.” Miller-El v. Cockrell, 537 U.S. 322, 340, 123 S.Ct. 1029, 154 L.Ed.2d 931 (2003) (Miller-El I). The state court’s factual findings are presumed to be sound unless the petitioner rebuts the “presumption of correctness by clear and convincing evidence.” 28 U.S.C. § 2254(e)(1). “The standard is demanding but not insatiable; ... ‘[djeferenee does not by definition preclude relief.’ ” Miller-El II, 545 U.S. at 240, 125 S.Ct. 2317 (quoting Miller-El I, 537 U.S. at 340, 123 S.Ct. 1029).
We recognize some ambiguity in the district court’s opinion about whether it applied the deference that section 2254(d)(2) requires. It first states that because of the legal error it found under section (d)(1), no AEDPA deference to factual findings was required. In the next breath, however, it recognizes that Cham-berlin “must demonstrate that the state court’s factual findings were unreasonable in light of the evidence presented” and cited Miller-El II’s use of the demanding “clear and convincing” standard required to overcome state court findings.6 We need not resolve this ambiguity because we can affirm on any ground supported by the record below. Dorsey v. Stephens, 720 F.3d 309, 314 (5th Cir. 2013). Applying AED-PA’s deferential standard, we conclude that the state court’s rejection of the Bat-son claim was based on an unreasonable determination of-the facts.
The Supreme Court and this court have both granted writs based on findings that state courts had made unreasonable factual determinations in rejecting Batson claims. Miller-El II, 545 U.S. at 266, 125 S.Ct. 2317; Reed, 555 F.3d at 382. In doing so, those cases relied heavily on comparative juror analysis. That analysis comes into play in the final stage of the Batson inquiry for determining whether a prosecutor used a peremptory strike in a racial[664]*664ly discriminatory manner. Before that point, the defendant must have first made a prima facie showing that the prosecutor exercised peremptory challenges on the basis of race; in response, the prosecutor must have articulated a race-neutral reason for striking the juror in question. Bat-son, 476 U.S. at 96-98, 106 S.Ct. 1712. That requires the court to then make the ultimate determination whether the defendant carried her burden of proving purposeful discrimination. Batson, 476 U.S. at 96-98, 106 S.Ct. 1712; see Purkett v. Elem, 514 U.S. 765, 768, 115 S.Ct. 1769, 131 L.Ed.2d 834 (1995).
“When the process reaches this step, the defendant may rely on all relevant circumstances to raise an inference of purposeful discrimination.” Fields v. Thaler, 588 F.3d 270, 274 (5th Cir. 2009) (quoting Miller-El II, 545 U.S. at 240, 125 S.Ct. 2317) (internal quotation marks omitted). The pattern of strikes here, while not dis-positive, is compelling evidence of intentional discrimination. See Miller-El I, 537 U.S. at 342, 123 S.Ct. 1029; Hayes v. Thaler, 361 Fed.Appx. 563, 570 (5th Cir. 2010). The State struck nearly two times as many black jurors as it accepted (eight strikes compared to five accepted, including one alternate), while accepting more than four times as many white jurors as it struck (five strikes compared to twenty-three accepted, including three alternates). It exercised 62% of its strikes on black jurors, despite black jurors making up only 31% of qualified prospective jurors.
In other words, black jurors were more than three times more likely to be struck by the prosecutor than white jurors. “Happenstance is unlikely to produce this disparity.” Miller-El II, 545 U.S. at 241, 125 S.Ct. 2317 (quoting Miller-El I, at 342, 123 S.Ct. 1029); Batson, 476 U.S. at 93, 106 S.Ct. 1712 (noting that “seriously disproportionate exclusion” of black jurors “is itself such an ‘unequal application of the law ... as to show intentional discrimination’ ”).
The dissent disagrees with the conclusion we draw from these statistics, but does not call into question the fundamental point that the prosecutor was far more likely to strike black potential jurors than whites.7 To distract from that disparity, the dissent compares the number of black jurors the prosecutor accepted to the number of spots on the jury.8 This is not [665]*665an illuminating comparison. To determine the prosecutor’s pattern of strikes, it is most probative to compare apples to apples: the number of black jurors the prosecutor accepted to the total number of jurors he accepted. He did not accept twelve total jurors. He accepted twenty-four, less than 17% of whom were black. The mismatch between the metrics the dissent compares is apparent when applied to white prospective jurors. The government accepted twenty white jurors during jury selection, enough to compose a nonsensical 167% of the jury.
The sequence of the strikes is also telling. The State used the vast majority of its early strikes against black jurors (seven of its first nine, including its sixth against Sturgis and its eighth against Minor) and only later — after defense counsel’s repeated objections and when it was running out of strikes—accepted the two black jurors who ended up on the jury (the second in a moment of confusion when the prosecutor believed the juror had already been struck). See Miller-El II, 545 U.S. at 249-50, 125 S.Ct. 2317 (finding unpersuasive that, towards the end of jury selection, the prosecution accepted a black juror, noting that most of the prosecution’s challenges were gone and the prosecutor “had to exercise prudent restraint” at that point).
But “the critical question in determining whether a prisoner has proved purposeful discrimination at step three is the persuasiveness of the prosecutor’s justification for his peremptory strike.” Miller-El I, 537 U.S. at 338-39, 123 S.Ct. 1029. To determine whether a prosecutor’s reason is persuasive, courts consider whether it was applied in a race-neutral way: “If a prosecutor’s proffered reason for striking a black panelist applies just as well to an otherwise-similar non-black who is permitted to serve, that is evidence tending to prove purposeful discrimination to be considered at Batson’s third step.” Miller-El II, 545 U.S. at 240-41, 125 S.Ct. 2317. Such “side-by-side comparisons,” also common in employment discrimination cases,9 are often “[m]ore powerful” than bare statistics. Miller-El II, 545 U.S. at 241, 125 S.Ct. 2317.
This is the comparative juror analysis we have mentioned. It was used in Miller-El II, in which the state struck a potential black juror purportedly because he “said that he could only give death if he thought a person could not be rehabilitated.” 545 U.S. at 243, 125 S.Ct. 2317. If that were the real reason, the Court noted, the prosecutor “should have worried about a number of white panel members he accepted” who expressed similar views. Id. at 244-45, 125 S.Ct. 2317. Likewise, although the prosecutor’s purported reason for striking another prospective juror (that he considered death “an easy way out”) was reasonable on its face, “its plausibility [wa]s severely undercut by the prosecution’s failure to object to other panel members who expressed views much like [his.]” Id. at 247-48, 125 S.Ct. 2317; see also Foster v. Chatman, — U.S. -, 136 S.Ct. 1737, 1751, 195 L.Ed.2d 1 (2016) (finding “otherwise legitimate reason[s]” for striking prospective black jurors “difficult to credit in light of the State’s acceptance of’ white jurors to whom those reasons also applied); Snyder v. La., 552 U.S. 472, 483, [666]*666128 S.Ct. 1203, 170 L.Ed.2d 175 (2008) (same); Reed, 555 F.3d at 372-73 (same).
In Miller-El II, the Court found the prosecutor’s reasons to be pretextual and thus powerful evidence of discrimination, even though other reasons the prosecutor gave for striking black jurors did not also apply to accepted white jurors. 545 U.S. at 247, 125 S.Ct. 2317. For example, the prosecutor gave an additional reason for striking two black jurors — that they had relatives who had been convicted of a crime— which did not apply to the white jurors to whom the Court compared them. Miller-El II, 545 U.S. at 246-47, 125 S.Ct. 2317; id. at 290-91, 125 S.Ct. 2317 (Thomas, J., dissenting). The Court nonetheless rejected the argument that pretext can be found only when an accepted white juror “mateh[es] all” of the reasons the prosecutor gave for striking a black juror. Id. at 247 n.6, 125 S.Ct. 2317 (quoting Thomas, J., dissenting). A rule that “no comparison is probative unless the situation of the individuals compared is identical in all respects” identified by the prosecutor would, it explained, “leave Batson inoperable; potential jurors are not products of a set of cookie cutters.”10 Id.
The jurors “identical in all respects” that Miller-El II thought unlikely exist here. Every reason the prosecutor identified for excluding Sturgis and Minor applied to Cooper, the white juror who was not struck.11 All three said they were “not sure” if they were emotionally capable of announcing a verdict of death; were “not sure” if they would hold the State to a higher burden of proof than the law requires given that it was a death penalty case; and “yes,” they would want to be one hundred percent certain of the defendant’s guilt before finding her guilty. Comparative juror analysis thus shows that the prosecutor’s reason for striking Sturgis and Minor could not have been their an-swérs to questions 30, 34, and 35. Otherwise, he would not have accepted Cooper who had the same answers the prosecution did not like. The perfect match among the answers of these jurors means that even more than in the other cases that have found pretext based on a comparative juror analysis, “[t]he prosecutors’ chosen race-neutral reasons for the strikes do not hold up and are so far at odds with the evidence that pretext is the fair conclusion, indicating the very discrimination the explanations were meant to deny.” Miller-El II, 545 U.S. at 265, 125 S.Ct. 2317; Snyder, 552 U.S. at 485, 128 S.Ct. 1203 (“The prosecution’s proffer of this pretextual explanation naturally gives rise to an inference of discriminatory intent.”); Reed, 555 F.3d at 380-81 (“[T]he comparative analysis demonstrates what was really going on: the prosecution used its peremptory challenges to ensure that African-Americans would not serve.”).
The State does not contest the obvious — that on the questions the prose-[667]*667eutor cited during jury selection as his reasons for excluding Sturgis and Minor, Cooper gave the same responses. Instead, it argues that it should now be able to identify differences among those prospective jurors on their responses to other questions. One example is the three prospective jurors’ differing answers to a different question about the death penalty (question 53): Cooper was strongly in favor of the death penalty whereas Sturgis “generally favored” it and Minor had “no opinion.” The State and dissent urge us to accept this justification. The problem is that Miller-El II rejected prosecutors’ ability to justify their strikes based on reasons not offered during jury selection and appellate courts’ ability to come up with new rationales on prosecutors’ behalf:
It is true that peremptories are often the subject of instinct and it can sometimes be hard to say what the reason is. But when illegitimate grounds like race are an issue, a prosecutor simply has got to state his reasons as best he can and stand or fall on the plausibility of the reasons he gives. A Batson challenge does not call for a mere exercise in thinking up any rational basis. If the stated reason does not hold up, its pre-textual significance does not fade because a trial judge, or an appeals court, can imagine a reason that might not have been shown up as false.
545 U.S. at 252, 125 S.Ct. 2317; see also Reed, 555 F.3d at 376 (“We must consider only the State’s asserted reasons for striking the black jurors and compare those reasons with its treatment of the nonblack jurors” (citing Miller-El II, 545 U.S. at 252, 125 S.Ct. 2317)).
Despite this unequivocal command, the dissent argues we can nonetheless consider the jurors’ views on questions not cited by the prosecutor after he was asked to justify the strikes. It first says we should do so because Miller-El II instructed courts to evaluate whether a prosecutors’ stated reason is plausible “in light of all evidence with a bearing on it.” 545 U.S. at 251-52, 125 S.Ct. 2317. But that should not be read to provide an end run around the prohibition on considering new reasons set forth in the same opinion. Miller-El II shows the difference between evidence bearing on plausibility, which reviewing courts should consider, and new reasons, which they may not. In evaluating whether proffered reasons were plausible, Miller-El II looked to evidence of the prosecutor’s veracity: did he rely on misrepresentations about stricken jurors’ answers, accept jurors with similar answers to stricken jurors, or give inconsistent explanations for strikes? Id. at 244-
51, 125 S.Ct. 2317.
In contrast, Miller-El II would not consider a new reason this court identified on appeal. Id. at 252, 125 S.Ct. 2317. The prosecutor initially had explained a strike by saying the potential juror thought the death penalty was “too easy on some defendants.” Id. at 250, 125 S.Ct. 2317. When the defendant pointed out during federal habeas that the same reason applied to white jurors the state accepted, this court found the real reason for the strike must have been the struck black juror’s “general ambivalence about the [death] penalty and his ability to impose it.” Id. at 248-51, 125 S.Ct. 2317. Miller-El II rejected this approach, similar to that of the dissent, because the “Court of Appeals’s ... substitution of a reason ... does nothing to satisfy the prosecutor’s burden.” Id. at 252, 125 S.Ct. 2317.
Other circuits conducting comparative jury analysis have also read Miller-El II as requiring that the “validity of a strike challenged under Batson must ‘stand or fall’ on the plausibility of the explanation given for it at the time, not new post hoc [668]*668justifications.” United States v. Taylor, 636 F.3d 901 (7th Cir. 2011); see also Love v. Cate, 449 Fed.Appx. 570, 572 (9th Cir. 2011) (refusing to consider the State’s post-trial explanation that white jurors it accepted “had non-racial characteristics that distinguished them from the black venire-member” the State struck because “the prosecutor never stated to the state trial court that he relied on these characteristics, even though Batson required him to articulate his reasons”); McGahee v. Alabama Dep’t Of Corr., 560 F.3d 1252, 1269 (11th Cir. 2009) (faulting the state appellate court for bolstering the prosecutor’s reason with a new explanation when the “State never offered such a full explanation”). In Taylor, the only reason the prosecutor gave during jury selection for a strike was that the black juror was unwilling to impose the death penalty on a non-shooter, a position also taken by accepted white jurors. 636 F.3d at 903, 905. After a remand because of concerns about the strike, the district court credited a different justification: the jurors’ differing views about the death penalty — which “the prosecutor did not say a word about” at trial— explained their disparate treatment. Id. at 905-06. In the opinion reversing, Judge Sykes explained that it was clear error to accept “new, unrelated reasons extending well beyond the prosecutor’s original justification.” Id. at 906.
The dissent thinks this prohibition on post-trial justifications can be overcome by repackaging the argument made by the State about the different answers to question 53. What the State candidly recognized is a new reason for striking the black jurors is now a new reason for keeping the white juror. The dissent presumes the prosecution kept Cooper because of his answer to question 53, as it “might have alleviated the prosecution’s concerns regarding his answers to questions 30, 34, and 35” that were identical to those of the struck black jurors. Dissent at 672. Of course, this is just the other side of the same coin. If the difference between the three was question 53, that would mean Sturgis and Minor were struck not only because of their answers to questions 30, 34, and 35, but also because of their more lukewarm support of the death penalty conveyed in response to question 53. As the “comparative juror analysis” name indicates, the inquiry is a comparative one that requires differentiating the answers of struck and accepted jurors. That means citing different answers to the same question as a reason for keeping one juror is the same as saying the difference was a reason for striking the other juror. To use a simple example, assume a prosecutor struck Jurors A and B on the ground that they wore hats in the courtroom. But Juror C was also wearing a hat. When this is later pointed out, the court speculates that Juror C must have been kept in the panel despite the hat because she expressed greater support for the death penalty on a questionnaire than did Jurors A and B. That would mean that the hat was not the dealbreaker; it alone was not enough for a strike as shown by the acceptance of Juror C. Jurors A and B thus would have been struck, per the court’s conjecture, because they wore a hat and were less supportive of the death penalty. And if that were the case, Miller-El II says the prosecutor should have cited both of those reasons.
The dissent’s position that courts may credit new reasons jurors were kept despite sharing the trait the prosecution claimed justified striking black jurors — a position for which it cites no authority— would make Miller-El ITs bar on considering new reasons for strikes meaningless. Take Miller-El II itself. The new reason for striking the black juror our court offered that the Supreme Court rejected— his ambivalence about the death penalty— [669]*669could just as easily have been treated as a reason for keeping the white jurors: their firmer support of the death penalty. 545 U.S. at 250-52, 125 S.Ct. 2317. Indeed, that is how the Miller-El II dissent characterized the difference. Id. at 289, 125 S.Ct. 2317 (Thomas, J., dissenting). The Miller-El II majority’s refusal to consider that new justification, whether framed as a reason for excluding the black juror or in opposite terms as a reason for keeping the white jurors, binds us.
The State argues that Miller-El II’s rule against after-the-fact justifications creates an unfair asymmetry in which it is held to the reasons it offered .at trial whereas the defendant can wait until the appeal to identify jurors like Cooper who have the same answers as people who were struck. Whatever the soundness of this complaint, it again is rejected by the leading decisions employing comparative juror analysis. See Miller-El II, 545 U.S. at 240, 125 S.Ct. 2317 (conducting comparative analysis on habeas review despite no such analysis being presented to state courts); Reed, 555 F.3d at 372-73 (same); Woodward, 580 F.3d at 338 (same); see also Smith v. Cain, 708 F.3d 628, 638 (5th Cir. 2013) (“[A]lthough Smith did not point to specific jurors for comparative analysis, we have conducted an in-depth review of the record.... ”). That Chamberlin’s counsel did not rebut the reasons presented by the prosecutor during jury selection is also beside the point. See Woodward, 580 F.3d at 338 (holding in a section 2254 review that a comparative analysis is appropriate even when defense counsel did not rebut the prosecutor’s stated reasons for striking jurors at trial) (citing Reed, 555 F.3d at 364))). These and the other arguments the State makes challenging the impact of comparative juror analysis also ignore that the approach is utilized only at the final stage of the Batson inquiry, which a trial court need not reach if it has properly found that the defendant did not establish a prima facie case of discrimination. And at that final stage, after the serious accusation of racial discrimination has been leveled and a prima facie case to support it found,12 it does not seem too much to ask prosecutors to list all the reasons justifying their strikes.
Although Miller-El II and Reed focus on comparative juror analysis in holding that the state courts’ factual finding of no Bat-son violation met AEDPA’s threshold of unreasonableness, those opinions also cite other evidence that supported that conclusion. In Miller-El II, the prosecutor had requested multiple jury shuffles, questioned jurors of different races inconsistently, and the Dallas district attorney’s office had a history of racial discrimination. Id. at 255-66, 125 S.Ct. 2317. Reed involved the same district attorney’s office and cited its history. 555 F.3d at 381-82; see also Foster, 136 S.Ct. at 1754 (stating that in addition to a comparative juror analysis “the shifting explanations, the misrepresentations of the record, and the persistent focus on race in the prosecution’s file” were evidence of discriminatory motive).
[670]*670But both our court (in an unpublished case) and another circuit have issued writs under AEDPA relying solely on comparative juror analysis to find a Batson violation. Hayes, 361 Fed.Appx. at 573; see also Drain v. Woods, 595 Fed.Appx. 558, 571-81 (6th Cir. 2014) (granting writ by finding a Batson violation relying only on a comparative juror analysis).13 That is not surprising as we have recognized since the early years of addressing Batson claims that the “decisive question will normally be whether a proffered race-neutral reason should be believed” because “there will seldom be any evidence that the claimant can introduce — beyond arguing that the explanations are not believable or pointing out that similar claims can be made about non-excluded jurors who are not minorities.” United States v. Bentley-Smith, 2 F.3d 1368, 1373-74 (5th Cir. 1993); cf. Reeves v. Sanderson Plumbing Prods., Inc., 530 U.S. 133, 147, 120 S.Ct. 2097, 147 L.Ed.2d 105 (2000) (explaining that general principles of evidence law allow a fact-finder to “reasonably infer from the falsity of the explanation that the employer is dissembling to cover up a discriminatory purpose”). When all of a prosecutor’s reasons are shown to be pretextual, it is hard to see how a court could reasonably find those reasons credible. See Miller-El I, 537 U.S. at 338-39, 123 S.Ct. 1029 (deeming Batson's “critical question” the credibility of the prosecutor’s reason).
In any event, there is more here than just the discrediting of the prosecutor’s explanation that comparative juror analysis compels us to find. Jury selection involved a definitive pattern of the prosecution striking black jurors, resulting in a stark disparity in the percentage of blacks struck as opposed to whites. And because a seated white juror gave identical answers to those cited in excluding the two black venire members, there is an absence of any nonpretextual justification for the strikes. Contrast Miller-El II, 545 U.S. at 246-47, 247 n.6, 125 S.Ct. 2317 (finding Batson violation even though the struck black venire member gave answers about rehabilitation and his brother’s criminal history that were not also provided by the accepted white jurors). Clear and convincing evidence, including more damning comparative juror analysis than existed in Miller-El II or Reed, thus rebuts the state court’s finding of no discrimination. It was unreasonable not to conclude that the strikes of Sturgis and Minor were “motivated in substantial part by discriminatory intent.” Snyder, 552 U.S. at 478, 485, 128 S.Ct. 1203.
The dissent argues that we are creating a system that allows adept prosecutors to avoid comparative juror analysis by giving vague and broad reasons for their strikes.' But in response to such reasons, Batson obligates the trial judge to require that the prosecutor give a “clear and reasonably specific” explanation for a challenge. 476 U.S. at 98 n.20, 106 S.Ct. 1712. More fundamentally, ease of evasion is not just a criticism of Miller-El IPs comparative juror analysis, it has been a basis for attacking the Batson framework from the beginning. See Batson, 476 U.S. at 106, 106 S.Ct. 1712 (Marshall, J., concurring); Rice v. Collins, 546 U.S. 333, 343, 126 S.Ct. 969, 163 L.Ed.2d 824 (2006) (Breyer, J., concurring). It would only exacerbate that problem to ignore Batson violations when the record reveals them.
[[Image here]]
[671]*671“Two peremptory strikes on the basis of race are two more than the Constitution allows.” Foster, 136 S.Ct. at 1755. Chamberlin is entitled to a new trial be.fore a jury selected without regard to race. The judgment of the district court is AFFIRMED