Before JONES, HIGGINSON, and OLDHAM, Circuit Judges. EDITH H. JONES, Circuit Judge: James Garfield Broadnax was convicted of capital murder and sentenced to death for robbing and fatally shooting two men. After exhausting his state remedies, Broadnax filed a federal habeas petition under 28 U.S.C. § 2254. The district court rejected his petition and denied a certificate of appealability (“COA”). Broadnax sought a COA under 28 U.S.C. § 2253(c)(2) to appeal numerous issues. We granted a COA and received additional briefing on a single issue pertinent to his Batson challenges to the jury’s makeup: “Whether the district court erroneously concluded that the spreadsheet was barred by Pinholster and 28 U.S.C. § 2254(d)(2).” Broadnax v. Davis, 813 F. App’x 166 (5th Cir. 2020) (per curiam). We now AFFIRM the district court’s refusal to consider newly discovered evidence relevant to Broadnax’s Batson claim because Cullen v. Pinholster, 563 U.S. 170, 131 S. Ct. 1388 (2011), bars its consideration. We also explain why COA is DENIED on Broadnax’s other claims. BACKGROUND On June 19, 2008, Broadnax and Demarius Cummings fatally shot and robbed Stephen Swan and Matthew Butler in Garland, Texas. Broadnax was convicted of capital murder and sentenced to death. After his arrest, Broadnax gave several interviews with local news stations which became the foundation of the State’s case at trial. In them, he confessed to the murder and robbery and provided explicit details of the murder. He admitted that he alone killed Swan and Butler, that he had no remorse, and he hoped for the death penalty. During voir dire, Broadnax challenged the prosecution’s use of peremptory strikes to remove all prospective black jurors and a Hispanic juror based on Batson v. Kentucky, 476 U.S. 79, 106 S. Ct. 1712 (1986).[1] The trial court initially denied all Broadnax’s challenges but eventually reseated one of the struck jurors. At trial, Broadnax did not dispute that he killed the victims, but he developed an extensive mitigation case that focused on his drug use at the time of the offenses. Broadnax presented expert testimony to the effect that because he committed the crimes at the age of nineteen, his brain would not have been fully developed. Dr. Frank Lane, a jail physician who treated Broadnax, testified that Broadnax claimed he was hallucinating, was paranoid, and did not remember talking to the media. Broadnax also told him that he had used PCP at the time of the offense. Because of this, Dr. Lane opined that Broadnax was most likely suffering from mood and perceptual disturbances due to prior PCP use. On direct appeal, Broadnax raised fifty-six points of error, including his challenges to the trial court’s Batson rulings. The Texas Court of Criminal Appeals (“TCCA”) affirmed the conviction and sentence. Broadnax v. State, AP-76,207, 2011 WL 6225399 (Tex. Crim. App. Dec. 14, 2011), cert. denied, 568 U.S. 828 (2012). Broadnax then filed his state habeas corpus petition. After an evidentiary hearing, the trial court recommended denial of relief and the TCCA adopted the trial court’s findings and conclusions. Ex parte Broadnax, WR-81,573-01, 2015 WL 2452758 (Tex. Crim. App. May 20, 2015), cert. denied, 136 S. Ct. 77 (2015). Having exhausted state remedies, Broadnax petitioned for federal habeas relief claiming ineffective assistance of counsel, Batson violations, erroneous evidentiary rulings, and errors in the punishment phase jury charge. He also challenged the constitutionality of the Texas capital punishment scheme and the death penalty. As part of his Batson challenges, Broadnax submitted for the first time a spreadsheet created by the Dallas County District Attorney’s Office in preparation for voir dire. The spreadsheet specified the race and gender of the veniremembers and bolded the names of prospective black jurors. As Broadnax admits, this document was previously withheld by the District Attorney’s Office as privileged work product, and he only gained access when the office revised its policy. The spreadsheet was not part of the record before the state court. The district court refused to consider the spreadsheet because in Pinholster, the Supreme Court “bar[red] [the court] from considering new evidence that was not properly before the Texas Court of Criminal Appeals when it rejected Broadnax’s Batson claims on direct appeal.”[2] Subsequently, the district court denied habeas relief on all grounds and further denied a COA on all claims. Broadnax appealed to this court and moved for a COA. This court granted a COA for one issue: “Whether the district court erroneously concluded that the spreadsheet was barred by Pinholster and 28 U.S.C. § 2254(d)(2).” Broadnax v. Davis, 813 F. App’x 166 (5th Cir. 2020) (per curiam). We first address the Pinholster/Batson claim and then the denial of COA on Broadnax’s other issues. STANDARD OF REVIEW Federal habeas proceedings are governed by the Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”), and a petitioner must first obtain a COA before he may appeal the district court’s denial of habeas relief. See 28 U.S.C. § 2253(c)(1)(A); Miller-El v. Cockrell, 537 U.S. 322, 335–36, 123 S. Ct. 1029, 1039 (2003) (“Miller-El I“). To obtain a COA, a petitioner must make “a substantial showing of the denial of a constitutional right.” 28 U.S.C. § 2253(c)(2). This standard requires the petitioner to demonstrate “that reasonable jurists could debate whether (or, for that matter, agree that) the petition should have been resolved in a different manner or that the issues presented were adequate to deserve encouragement to proceed further.” Miller–El I, 537 U.S. at 336, 123 S. Ct. at 1039 (internal quotes omitted). “[A]ny doubt as to whether a COA should issue in a death-penalty case must be resolved in favor of the petitioner.” Pippin v. Dretke, 434 F.3d 782, 787 (5th Cir. 2005). When considering a COA application, this court has jurisdiction to determine only whether a COA should issue, not the ultimate merits of the petitioner’s claims. Ward v. Stephens, 777 F.3d 250, 255 (5th Cir. 2015). “This threshold inquiry does not require full consideration of the factual or legal bases adduced in support of the claims. In fact, the statute forbids it.” Miller-El I, 537 U.S. at 336, 123 S. Ct. at 1039. Indeed, “[t]he question is the debatability of the underlying constitutional claim, not the resolution of that debate,” and a “claim can be debatable even though every jurist of reason might agree, after the COA has been granted and the case has received full consideration, that petitioner will not prevail.” Id. at 338, 342, 123 S. Ct. at 1040, 1042. If a COA is granted, our jurisdiction extends only to “the issue specified in the COA.” Simmons v. Epps, 654 F.3d 526, 535 (5th Cir. 2011) (per curiam). “In a habeas corpus appeal, we review the district court’s finding of fact for clear error and its conclusions of law de novo.” Higgins v. Cain, 720 F.3d 255, 260 (5th Cir. 2013). AEDPA bars habeas relief for a claim adjudicated on the merits by a state court absent a showing that the decision was either (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.” 28 U.S.C. § 2254(d). A state court decision is “contrary to” federal law if, on materially indistinguishable facts, it reaches a conclusion opposite of a Supreme Court case. Williams v. Taylor, 529 U.S. 362, 413, 120 S. Ct. 1495, 1523 (2000). The decision is an unreasonable application of federal law if the state court correctly identified the governing legal principle but unreasonably applied it to the facts of the case. Id. Factual findings by the state court are presumed correct unless rebutted “by clear and convincing evidence.” 28 U.S. § 2254(e)(1). PINHOLSTER , the JUROR SPREADSHEET, and the BATSON CLAIM This court granted a COA in order to consider more fully the contention that the federal district court should have admitted the racially annotated juror spreadsheet, not made available in the state courts, when it evaluated Broadnax’s Batson claim. In Pinholster, the Supreme Court considered “whether review under [AEDPA] § 2254(d)(1) permits consideration of evidence introduced in an evidentiary hearing before the federal habeas court.” 563 U.S. at 180, 131 S. Ct. at 1398. This provision limits federal court habeas review to claims that were “adjudicated on the merits” in state courts. For such claims, the Court concluded that “the record under review is limited to the record in existence at that same time i.e., the record before the state court.” Id. at 182, 131 S. Ct. at 1398. Further, the Court noted, § 2254(d)(2) provides “additional clarity” by expressly confining review of the underlying evidence to “evidence presented in the State court proceeding.” Id. at 185 n.7, 131 S. Ct. at 1400 n.7. The point of AEDPA, the Court explained, is to require prisoners first to exhaust state court remedies before seeking federal relief, and “[i]t would be contrary to that purpose to allow a petitioner to overcome an adverse state- court decision with new evidence introduced in a federal habeas court and reviewed by that court in the first instance effectively de novo.” Id. at 182, 131 S. Ct. at 1399. The Court added that, “[a]lthough state prisoners may sometimes submit new evidence in federal court, AEDPA’s statutory scheme is designed to strongly discourage them from doing so.” Id. at 186, 131 S. Ct. at 1401. Pinholster thus confirms limitations on a federal habeas court’s consideration of new evidence when reviewing claims that have been adjudicated on the merits in state court.[3] In such circumstances, the petitioner must demonstrate that habeas relief is warranted under § 2254(d) on the state court record alone. If the petitioner succeeds in satisfying this threshold requirement, then a federal habeas court may entertain new evidence pursuant to the limitations of § 2254(e)(2). See, e.g., Smith v. Cain, 708 F.3d 628, 634–35 (5th Cir. 2013) (holding that an evidentiary hearing was permissible after the federal trial court first determined “on the basis of the state court record that the state court’s Batson analysis was contrary to, or at least involved an unreasonable application of, clearly established Federal law” (internal quotations omitted)). Broadnax principally asserts that a narrow exception to Pinholster benefits him here. Alternatively, he claims the spreadsheet is admissible under the approach of Smith v. Cain. We disagree with both contentions. The Pinholster-Exception Theory While admitting that he raised Batson challenges to the prosecutors’ peremptory strikes of minority jurors and exhausted that claim in state court, Broadnax alleges that the spreadsheet “fundamentally alters” his Batson claim because it was “withheld” by the prosecution and was made unavailable to him in state proceedings. He relies on a footnote in Pinholster, which recognized that in some instances new evidence may present a new claim of which federal habeas courts may take cognizance. Pinholster, 531 U.S. at 186 n.10, 131 S. Ct. at 1401 n.10 (stating that the dissent’s “hypothetical involving new evidence of withheld exculpatory witness statements . . . may well present a new claim”). But the Court declined in that footnote “to draw the line between new claims and claims adjudicated on the merits.” Id. Broadnax asserts that under this exception-to-Pinholster, the “withheld” spreadsheet constitutes such “new evidence” that “fundamentally alters” his Batson claims and should have been added to the district court’s analysis. This theory raises a difficult question at the outset. Despite repeated insinuations, Broadnax does not allege that the state improperly withheld the juror spreadsheet during state court proceedings. Indeed, Broadnax’s initial federal habeas attorney admitted that the spreadsheet was nondiscoverable attorney work product that was only disclosed to the defense, after the federal habeas proceeding commenced, because of a change in the District Attorney’s policies. The spreadsheet does not pertain to Broadnax’s guilt or innocence and therefore fell outside the prosecution’s Brady disclosure obligations. In contrast, Pinholster‘s majority footnote considers a hypothetical framed around exculpatory evidence and a potential Brady violation. Logically, more than one habeas “claim” could be predicated on distinct failures to disclose exculpatory or impeaching information pursuant to Brady. Broadnax ignores the distinction between the separability of Brady claims and the mere (alleged) evidentiary enhancement of a singular Batson claim by the introduction of the spreadsheet. Broadnax’s “new claim” is not just non-exculpatory but does not support any kind of freestanding Batson “claim” at all.[4] Whatever lines might be drawn pursuant to the Pinholster footnote, they are not implicated here. An additional impediment to the Pinholster exception theory is that this court’s post-Pinholster precedent offers no support for it. Within a year after the Supreme Court’s decision, this court held that “[t]he import of Pinholster is clear: because Lewis’s claims have already been adjudicated on the merits, § 2254 limits our review to the record that was before the state court.” Lewis v. Thaler, 701 F.3d 783, 791 (5th Cir. 2012). In so doing, the court refused to consider expert mitigation evidence offered for the first time in federal habeas. Another decision rejected the notion that, after Pinholster, the federal court should consider whether newly offered mitigation evidence constituted an unexhausted claim or “simply supplement[ed]” petitioner’s state court claim. Clark v. Thaler, 673 F.3d 410, 417 (5th Cir. 2012). Instead, the court stated, “[w]e consider only the record that was before the state habeas court.” Id. In yet another case, this court rejected the parties’ joint position on appeal that newly offered mitigation evidence concerning mental illness rendered a petitioner’s federal habeas claim unexhausted.[5] Ward, 777 F.3d at 258. Then, in addressing the merits, the court cited Pinholster and analyzed the § 2254(d)(1) claim in light of the state court record alone. Ward, 777 F.3d at 264. Notably, Smith is this court’s sole foray into applying Pinholster to a Batson claim. And when this court applied Pinholster, it reviewed only the state court record for its conclusion that the state courts decided a Batson claim “unreasonably” pursuant to § 2254(d)(1). Smith, 708 F.3d at 634–35. As a result, only then did the Smith court take into consideration additional evidence offered in federal court. Id. The Fifth Circuit cases cited by Broadnax, when carefully read, do not challenge this court’s otherwise uniform adherence to Pinholster. One of these granted a COA on an ineffectiveness claim. First and most important, a COA is not a definitive ruling on the merits of an issue in habeas.[6] See Nelson v. Davis, 952 F.3d 651, 671–72 (5th Cir. 2020). Nelson is not a binding statement of law on the issue garnering a COA. Second, the COA issue in Nelson concerned exhaustion. Citing Pinholster’s footnote 10 (refusing to draw lines between new claims and claims unadjudicated in state courts), Nelson stated that, “while ‘merely putting a claim in a stronger evidentiary posture is not enough,’ new evidence that ‘fundamentally alters the legal claim’ or ‘place[s] the claim in a significantly different legal posture’ can render it a new claim that was not adjudicated on the merits by the state court.” Id. (quoting the discussion of exhaustion in Ward, 777 F.3d at 258– 59). For COA purposes, the court declared, the issue was that “reasonable jurists could debate whether Nelson’s [ineffective assistance]-Participation allegations ‘fundamentally alter’ his [counsel ineffectiveness] claim, and so constitute a different and unexhausted claim.” Id. at 672. Two factually distinct theories of ineffective assistance were at least theoretically implicated. How this court might rule on the ultimate COA issue and a number of intertwined issues in Nelson is unknown at this time. Most important, Broadnax has never argued that his spreadsheet evidence represents an unexhausted claim. Escamilla v. Stephens, 749 F.3d 380, 394–95 (5th Cir. 2014), also fails to support Broadnax. That case held forthrightly that where a petitioner’s habeas counsel had raised an issue in the state habeas court, albeit ineffectively from a constitutional standpoint, the petitioner was barred by Pinholster from offering new evidence in federal court precisely because the original claim had been “fully adjudicated on the merits” in state court. Id. Contrary to Broadnax’s view, in light of Escamilla’s clear and unequivocal holding,[7] this court did not find that an exception to Pinholster exists whenever newly offered evidence “fundamentally alters” a claim previously presented to the state courts. The court merely stated that the additional evidence submitted to the federal habeas court in that case effected no such alteration. Id. at 395. Broadnax’s out-of-circuit citations are also unhelpful. He cites Wolfe v. Clark, but that case does not even cite, much less apply Pinholster. For that reason alone, Wolfe is inapposite. Wolfe held that a long-concealed police report, first offered in federal court, supported a new and separable Brady claim that had not been adjudicated in state court, rendering § 2254(d) irrelevant. 691 F.3d 410, 423 (4th Cir. 2012). Like Nelson, this aspect of Wolfe concerns the exhaustion doctrine. The spreadsheet fails to rise to that level, and the Batson claim raised here was adjudicated in state courts. Broadnax made a detailed showing in state court by pointing out alleged inconsistencies in the treatment of jurors as well as the final make-up of the empaneled jury. The spreadsheet offered in federal court reflects the prosecutors’ awareness of the race of prospective jurors. But it is not the “single, plainly momentous item of suppressed . . . evidence” for which habeas relief is warranted. Wolfe, 691 F.3d at 417. Thus, Broadnax’s Batson claim remains subject to the limitations of § 2254(d).[8] Finally, even if we were to accept the Pinholster-exception theory espoused by Broadnax, his argument would fail. The spreadsheet, at most, places Broadnax’s Batson claim “in a stronger evidentiary position;” in no way does it “fundamentally alter” the preexisting claim. As the district court noted, the spreadsheet “does nothing more than indicate that the Dallas County District Attorney’s Office made a point of memorializing the ethnicity and gender of the remaining members of the jury venire prior to the exercise of its peremptory challenges.” Broadnax v. Davis, No. 3:15-CV- 1758-N, 2019 WL 3302840, at *43 n.73 (N.D. Tex. July 23, 2019). Batson claims are evaluated under a three-step process: (1) the defendant makes a prima facie showing that the peremptory challenge was based on race; (2) the prosecution provides a race-neutral basis for the strike; (3) the trial court determines whether the prosecutor purposefully discriminated against the juror. Foster v. Chatman, 136 S. Ct. 1737, 1747 (2016). The spreadsheet arguably enhances Broadnax’s argument at the first step, and it may be relevant to the third. But the prosecution was still required to—and did— provide racially neutral reasons for each of the strikes. The spreadsheet alone is no smoking gun; it fails to render all those reasons merely pretextual. Moreover, the district court observed that the Dallas County District Attorney’s Office has twice been criticized by the United States Supreme Court for the use of racially discriminatory peremptory strikes.[9] Broadnax, 2019 WL 3302840, at *43 n.73. The office would have had considerable motivation to identify which jury venire members belonged to a protected class when preparing to defend its use of peremptory challenges.[10] For all these reasons, we reject the theory that a narrow exception articulated in footnote 10 of Pinholster’s majority opinion required the federal district court to admit and evaluate the District Attorney’s spreadsheet pertinent to Broadnax’s fully adjudicated and exhausted Batson claim. Smith v. Cain Approach Broadnax alternatively contends that the state court decision was “contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States” and was “based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding,” rendering Pinholster‘s prohibition on new evidence inapplicable. 28 U.S.C. § 2254(d); Smith, 708 F.3d at 634–35. This argument finds no traction in the record. Broadnax alleges that the State exercised its peremptory challenges in a racially discriminatory manner and intentionally struck every nonwhite veniremember. He marshals three arguments in support of this contention. First, Broadnax asserts that “the State treated white and nonwhite veniremembers differently.” Second, Broadnax raises as prima facie evidence of discrimination that the prosecution used its peremptory strikes to remove 100% of the nonwhite venire members. Finally, Broadnax argues that the trial court’s reinstatement of one struck African-American juror was not a sufficient remedy for the Batson violation. The district court conducted a lengthy analysis of the state’s contemporaneously expressed reasons for exercising strikes on each of the challenged jurors. In the end, it concluded that “Broadnax has failed to present this Court with clear and convincing evidence showing the state trial court’s implicit credibility findings (regarding the prosecution’s race-neutral reasons for its peremptory strikes against these venire members) were erroneous,” as required by 28 U.S.C. § 2254(e)(1). Broadnax, 2019 WL 3302840, at *43. Thus, the prior dismissal of the Batson claims was “neither contrary to, nor involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States, nor resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in his trial and direct appeal.” Id. For the following reasons, we agree. After a seventeen-page evaluation of the trial court record concerning the state’s peremptory strike on each of eight African-American veniremembers, one of whom was eventually seated, the district court found that the proffered justifications for the challenges “all constituted racially neutral, objectively verifiable, record-based, reasons for a prosecutorial peremptory strike.”[11] We need not repeat the district court’s exhaustive and convincing examination of each strike, but the strikes share common, race-neutral characteristics. The state gave each prospective juror a questionnaire. Two questions on the first page of the questionnaire are relevant to this issue. The first asked “Are you in favor of the death penalty?” The state struck every veniremember, regardless of race, who indicated he or she was not in favor of the death penalty, including five against whose strikes Broadnax’s counsel raised Batson objections. Each of the five gave various explanations for opposing the death penalty, ranging from a belief in second chances to concern about erroneous convictions. Second, the questionnaire asked veniremembers to circle one out of five possible responses to the following question: “With reference to the death penalty, which of the following statements best represents your feelings?” Option three stated: “Although I do not believe that the death penalty ever ought to be invoked, as long as the law provides for it, I could assess it under the proper set of circumstances.” Again, the state struck every veniremember, regardless of race, who selected option three. Broadnax’s counsel raised Batson objections to the strikes of two veniremembers who indicated this option. Broadnax counters that the state did not strike several white veniremembers who answered their questionnaires similarly to minority veniremembers who were struck. But with the exception of Juror Long, every Batson-challenged veniremember who was excluded from the jury indicated he or she was not in favor of the death penalty and/or believed the death penalty ought not be invoked. The state struck all who answered this way, a fact Broadnax glosses over. Moreover, while defendants need not demonstrate that white and nonwhite veniremembers were identical in all respects to demonstrate a Batson challenge, “the comparator-juror must be similar in the relevant characteristics.” Herbert v. Rogers, 890 F.3d 213, 223 (5th Cir. 2018). As the district court noted, given the extensive nature of the questionnaire, it is “hardly surprising—or conclusive of anything” that there would be similarities in some answers between struck and non-struck veniremembers. Broadnax, 2019 WL 3302840, at *43 n.73. Juror Long was the sole minority veniremember who expressed support for the death penalty and did not select option three. However, Long indicated that she would be “automatically prevented” from imposing the death penalty if the defendant was using drugs or alcohol at the time of the offense. As the state knew that intoxication would be a core component of the defense theory, Long’s answer was highly prejudicial to the state’s case. Moreover, several of Long’s explanations for her answers revealed mixed feelings about the death penalty. While one other veniremember considered intoxication to be a mitigating circumstance, Broadnax musters no other potential juror who believed that intoxication automatically rendered a defendant ineligible for the death penalty. Yet this automatic ineligibility formed the core of the state’s justification to the trial court for a peremptory strike. The district court correctly concluded that the state courts did not unreasonably apply Batson in rejecting this claim. Broadnax further urges that the state court erred in not properly considering several important circumstances in the state’s use of its peremptory strikes to remove 100% of the nonwhite veniremembers. Not so. As this court has previously emphasized, “[t]he Supreme Court has instructed that, when analyzing Batson challenges, ‘bare statistics’ are not the be-all end-all.” Chamberlin v. Fisher, 885 F.3d 832, 840 (5th Cir. 2018) (en banc) (quoting Miller-El v. Dretke, 545 U.S. 231, 241, 125 S. Ct. 2317, 2325 (2005) (Miller-El-II)). The “[m]ore powerful” evidence is a “side-by- side comparison[] of some black venire panelists who were struck and white panelists allowed to serve.” Miller-El-II, 545 U.S. at 241, 125 S. Ct. at 2325. The district court thoroughly conducted side-by-side analysis of the state courts’ determinations and correctly concluded that Batson was not unreasonably applied. Finally, Broadnax challenges the district court’s conclusion that the state courts did not unreasonably apply Batson in agreeing that reseating Juror Patterson was a sufficient response to the State’s Batson violation. After initially accepting the prosecution’s peremptory strike, the trial court found a Batson violation regarding Patterson and reseated him. Interestingly, the TCCA found on direct appeal that no Batson violation occurred. Broadnax, 2011 WL 6225399, at *4. Nevertheless, Broadnax proceeds as if a Batson violation occurred and disputes only the propriety of the remedy. He asserts that the trial court should have struck the entire jury panel and begun voir dire anew. The district court explained that no clearly established Supreme Court law requires dismissal of an entire jury panel in the face of a single Batson violation. Nor has Broadnax brought any such authority to the attention of this court. Batson itself expressly disavowed requiring trial courts to dismiss the entire panel and noted that reinstating the improperly challenged juror could be an adequate remedy. Batson, 476 U.S. at 99 n.24, 106 S. Ct. at 1725 n.24. While some other jurisdictions have suggested that dismissing the entire panel is the “better practice,” see, e.g., United States v. Walker, 490 F.3d 1282, 1295 (11th Cir. 2007), Texas views reinstating any excluded veniremember as an appropriate remedy for a Batson violation. See, e.g., State ex rel. Curry v. Bowman, 885 S.W.2d 421, 425 (Tex. Crim. App. 1993). In any event, the district court explained that “the new rule advocated by Broadnax in this federal habeas corpus proceeding is foreclosed by the nonretroactivity doctrine of Teague.” We find no error in the district court’s conclusions. Accordingly, because Broadnax did not surmount the standards embodied in § 2254(d), he had no basis to offer evidence outside the state court record, and the spreadsheet was correctly barred from consideration in federal court. PETITION FOR COA Broadnax advanced five other claims for relief in which he sought a COA. These contentions include: (1) an unqualified juror sat on his jury; the District Attorney’s decision to seek the death penalty was racially motivated; (3) Broadnax was unconstitutionally denied counsel when he gave media interviews; (4) his appellate counsel was ineffective for failing to challenge the admission of certain expert testimony on appeal; and (5) the district court applied erroneous legal standards and inadequately reviewed the record. Refusal of Strike for Cause Broadnax contends that reasonable jurists could debate the district court’s rejection of his claim that the state courts erroneously refused to disqualify juror John Vessels. Broadnax alleges that Vessels was unable to consider mitigation evidence. After reviewing Vessels’s responses, we conclude that the district court’s resolution of this issue is not debatable. A trial court must strike for cause a prospective juror who would automatically impose the death penalty without considering mitigating circumstances. Morgan v. Illinois, 504 U.S. 719, 729, 112 S. Ct. 2222, 2229 (1992). Despite this, the law does not oblige jurors to consider any specific circumstances as mitigating. See Soria v. Johnson, 207 F.3d 232, 244 (5th Cir. 2000). When asked during voir dire to give examples of mitigating evidence that would convince him to change a death sentence to a life sentence, Vessels responded he could not think of any. Additionally, Vessels expressed suspicion of evidence of intoxication and the defendant’s troubled upbringing as mitigating factors. The district court agreed with the TCCA that while Vessels considered these circumstances not to be mitigating, he would not absolutely refuse to consider mitigating evidence. As this court has explained, these types of evidence can be double-edged and may even be perceived as aggravating. Id.; see also Dorsey v. Quarterman, 494 F.3d 527, 533 (5th Cir. 2007) (“[T]he law is clear that a defendant in a capital case is not entitled to challenge prospective jurors for cause simply because they might view the evidence the defendant offers in mitigation of a death sentence as an aggravating rather than a mitigating factor.”). Vessels expressly stated that he could consider the mitigation special issue with an open mind and that he could answer “yes” on the issue. Thus, Vessels was not unwilling to put aside personal views, consider all the evidence, and follow the law; rather, he honestly acknowledged his suspicions concerning certain types of mitigating evidence. Reasonable jurists would not debate the district court’s determination that the state court reasonably rejected Broadnax’s claim that the trial court had to disqualify juror Vessels for cause. Selective Prosecution Broadnax raises a selective prosecution claim, arguing that the State sought to impose the death penalty on the basis of his race. The district court noted, without holding, that this claim was unexhausted, but it ruled on the merits instead. To succeed on his selective prosecution claim, Broadnax must overcome the presumption that a prosecutor acts in good faith and within his discretion; hence a defendant’s burden is to present clear evidence showing that the prosecutor’s decisions had both a discriminatory effect and a discriminatory motive or purpose. United States v. Armstrong, 517 U.S. 456, 465, 116 S. Ct. 1480, 1487 (2006); In re United States, 397 F.3d 274, 284 (5th Cir. 2005). To establish a racially discriminatory effect, a defendant must show that similarly situated individuals of a different race were not prosecuted. Armstrong, 517 U.S. at 465, 116 S. Ct. at 1486–87; In re United States, 397 F.3d at 284. The district court found that Broadnax is not “similarly situated” to any other offender of any race in Dallas County because of his media interviews. In the interviews, Broadnax confessed, described the crimes in graphic detail, repeatedly denied feelings of remorse, and demanded to receive the death penalty. What he said in the interviews “put Broadnax in a class by himself.” Broadnax argues this finding is debatable in light of a statistical study that allegedly shows the death penalty was imposed in Dallas County more often against African American defendants accused of victimizing whites than against other offenders and victims of other races. But such statistical evidence alone does not establish that “the decisionmakers in his case acted with discriminatory purpose.” McCleskey v. Kemp, 481 U.S. 279, 292, 107 S. Ct. 1756, 1767 (1987). Indeed, the Baldus study in McCleskey identified some stark statistical discrepancies. Id. at 287, 107 S. Ct. at 1764; Nevertheless, the Supreme Court denied relief. Further, the discretionary nature of the decision to seek a death sentence led the Court to caution against inferring discriminatory motive from statistical disparities alone. Id. at 297, 107 S. Ct. at 1770. Finally, Broadnax argues that the district court’s focus on Broadnax’s media interviews treats the “similarly situated” requirement too narrowly. We disagree. The prosecution relied on them extensively at trial, and the district court deemed their content sufficiently unusual that it declared Broadnax “sui generis.” The interviews graphically and voluntarily confessed Broadnax’s guilt, cravenness, and extreme future dangerousness. Moreover, Broadnax’s proffered view of “similarly situated,” by invoking broad generic commonalities such as racial characteristics and crimes charged, would render comparisons essentially meaningless. Thus, although in some circumstances there might be uncertainty about how to identify “similarly situated” offenders relevant to a selective prosecution claim, this is not such a case. The district court’s conclusion that Broadnax was not selectively prosecuted is not reasonably debatable. Uncounseled Media Interviews Broadnax next argues that he was denied his Sixth Amendment right to counsel while giving the media interviews, which he contends were “critical stages” of his trial. The interviews, which Broadnax voluntarily conducted, occurred after his initial appearance before the magistrate judge on June 21, 2008, but before he was appointed counsel on June 24. Before trial, Broadnax moved to suppress the interviews. The trial court denied the motion. Significantly, Broadnax signed the stations’ request forms seeking interviews, the reporters were not employed by law enforcement, and no law enforcement officer had requested that they conduct the interviews. An accused is guaranteed the right to counsel during all “critical stages” of his trial. Critical stages are “proceedings between an individual and agents of the State, whether ‘formal or informal, in court or out,’ that amount to ‘trial-like confrontations,’ at which counsel would help the accused ‘in coping with legal problems or . . . meeting his adversary.’” Rothgery v. Gillespie County, 554 U.S. 191, 212 n.16, 128 S. Ct. 2578, 2591 n.16 (2008) (cleaned up) (emphasis added). “[W]hat makes a stage critical is what shows the need for counsel’s presence.” Id. at 212, 128 S. Ct. at 2591. The district court concluded, pursuant to AEDPA, that the state courts, which examined numerous ways in which Broadnax sought to attack the admissibility of the interviews, did not unreasonably apply governing Supreme Court law nor unreasonably apply the facts to the legal standards, nor did they unreasonably determine the facts in light of the record. The court accordingly rejected his Fifth and Sixth Amendment claims. The district court alternatively ruled against Broadnax under a de novo standard. In so doing, the court examined whether the reporters who interviewed Broadnax were acting as agents of the State. Concluding they were not, the district court held that Broadnax’s Sixth Amendment claims lacked merit. In reaching its conclusion, the district court considered this circuit’s two-prong test for determining whether an informant was a government agent: whether the informant “(1) was promised, reasonably led to believe, or actually received a benefit in exchange for soliciting information from the defendant; and (2) acted pursuant to instructions from the State, or otherwise submitted to the State’s control.” Creel v. Johnson, 162 F.3d 385, 393 (5th Cir. 1998). There is no evidence in the record supporting either of these claims. The mere fact that reporters followed Sheriff’s Department procedures to request interviews does not prove that they submitted to the State’s control or received some benefit. The district court correctly observed that “[t]o hold otherwise would transform every media interview conducted with an individual under custodial detention into a custodial interrogation by a de facto state agent.” Not only does Broadnax lack evidence to support his “critical stage” assertion, but he cites no legal authority for the proposition that voluntary media interviews, conducted within days of an initial appearance, are a “critical stage” of a prosecution requiring the presence of defense counsel. The Teague principle, ensconced in AEDPA, forbids federal courts to make “new rules” of criminal procedure in habeas corpus review of final state convictions. Woods v. Donald, 575 U.S. 312, 317, 135 S. Ct. 1372, 1376–77 (2015); Teague v. Lane, 489 U.S. 288, 299–310, 109 S. Ct. 1060, 1069–1075 (1989). Reasonable jurists would not debate the district court’s conclusion that the state courts reasonably rejected this claim. Ineffective Assistance of Appellate Counsel Broadnax contends that his appellate counsel was ineffective for failing to challenge the admission of Dr. Price’s testimony on direct appeal. He alleges that although Dr Price, a rebuttal witness for the prosecution, did not render an expert diagnosis that Broadnax is a “psychopath,” his testimony concerning such a diagnosis was inflammatory, inadmissible and harmful under various provisions of Texas law. Despite this, Broadnax’s counsel did not raise the issue among nearly five dozen appellate issues he did assert. The district court acknowledged that Broadnax had not exhausted the claim in the state courts but rejected it on de novo review, concluding that the appellate counsel did not render ineffective assistance by failing to assert a meritless challenge to Dr. Price’s testimony. See Davila v. Davis, 137 S. Ct. 2058, 2067 (2017) (declining to raise a claim on appeal is not deficient performance unless that claim was plainly stronger than those actually presented to the appellate court). The district court found neither prong of the Strickland test for ineffective assistance of counsel satisfied. See Smith v. Robbins, 528 U.S. 259, 285, 125 S. Ct. 746, 764 (2000) (explaining that the Strickland test inquires whether counsel’s conduct was objectively unreasonable under then-current legal standards and whether counsel’s allegedly deficient performance “prejudiced” the petitioner). In addition, appellate counsel is not required to raise every non-frivolous claim on appeal. Id. at 288, 125 S. Ct. at 765–66. We need not review the professional quality of counsel’s appellate work under Strickland, because we cannot fault the district court’s conclusion that Broadnax failed to show that the admission of this evidence was harmful. On cross-examination, Dr. Price admitted “(1) psychopathy is not listed in the DSM-IV, (2) the closest thing to psychopathy in the DSM- IV is a personality disorder, (3) people who have the traits of psychopathy may not be a psychopath, and (4) some of the traits of a psychopath are consistent with those of an immature person.” Dr. Price admitted he was not making a mental health evaluation of Broadnax. When this testimony is viewed in light of the heinousness of the crime and Broadnax’s utter lack of remorse, even if its admission was erroneous under state law, the testimony was not prejudicial. Broadnax did not demonstrate that but for appellate counsel’s error, there was a reasonable likelihood that he would not have been sentenced to death. Strickland v. Washington, 466 U.S. 668, 687–91, 694, 104 S. Ct. 2052, 2064–66, 2068 (1984). The district court’s rejection of this ineffectiveness claim is not debatable by reasonable jurists. Relevant Legal Standards Lastly, Broadnax argues that the district court failed to apply the correct legal standards when reviewing his claims. This contention, at its core, is merely an attempt to dispute the court’s reasoning and portions of the evidence the district court considered in its monumental opinion. These assertions are meritless. First, the district court did not misapply AEDPA standards, because no change in the statutory standards relevant here was occasioned by Wilson v. Sellers, 138 S. Ct. 1188 (2018). The district court reviewed the state courts’ reasoning in accord with AEDPA. Second, the district court did not fail to base its conclusion about Broadnax’s selective prosecution claim on an independent review of the evidence in the record. Even a cursory reading of the district court’s opinion reveals that it carefully examined the record and new statistical evidence he raised. The district court did not consider the statistical evidence persuasive because Broadnax was not “similarly situated” to other defendants. Next, Broadnax challenges the district court’s finding of fact concerning when the State’s spreadsheet detailing the race and sex of veniremembers was created. That determination is irrelevant to the court’s conclusion, addressed above, that the spreadsheet was barred from consideration by Pinholster. Finally, Broadnax errs in claiming the district court made its own finding of fact that Broadnax was not under the influence of drugs at the time of his media interviews and thus ignored the testimony of mental health experts who examined Broadnax. This is inaccurate. The district court did consider testimony of each of these individuals concerning Broadnax’s drug use, even as it observed deficiencies in their statements. The district court also considered Broadnax’s demeanor during the taped interviews and the testimony of the jail employee who accompanied Broadnax to his interviews. CONCLUSION For these reasons, we have DENIED COA in part and AFFIRM the district court’s judgment in part.
