dissenting.
Because the prosecutor’s blatant homophobic hatemongering at sentencing has no place in the courtrooms of a civilized society, and Neill’s appellate counsel’s failure to raise the issue on direct appeal constitutes clear and plain prejudicial neglect, I respectfully dissent. Correctly or incorrectly, we have encapsulated the applicable Strickland jurisprudence into the term “dead-bang winner.” See United States v. Cook, 45 F.3d 388, 395 (1995) (citing Page v. United States, 884 F.2d 300, 302 (7th Cir.1989)).1 Whether we apply this more stringent standard, or an orthodox Strickland approach, we have before us a “dead-bang winner.”
While thinly disguising his intent by denying that a person’s “sexual preference” is an “aggravating circumstance,” the prosecutor deviously and despicably incited the jury with the following statement:
If I could ask each of you to disregard Jay Neill and take him out of the person but consider these things in a generic way. I want you to think briefly about the man you’re setting [sic] in judgment on ... and believe me, ... you have every thing in this case, the good, the bad, everything that the law allows to aid you in this decision. But just generic, just put in the back of your mind what if I was sitting in judgment on this person without relating it to Jay Neill, and I’d like to go through some things that to me depict the true person, what kind of person he is. He is a homosexual. The person you’re sitting in judgment on — disregard Jay Neill. You’re deciding life or death on a person that’s a vowed [sic] homosexual.... But these are areas you consider whenever you determine the type of person you’re setting [sic] in judgment on.... The individual’s homosexual. He’s in love with Robert Grady Johnson. •
(V Trial Tr. at 1285-86, 87.) I sternly reject the prosecutor’s disavowal and the “he brought it up” post hoc rationalization that this somehow justifies the use of hate as an appropriate adversarial tool.
Moreover, the record evidence calls into question whether Neill was sentenced to death by an impartial jury, which further erodes my confidence in the jury’s sentence. I would grant habeas relief and vacate Neill’s sentence.
*1200I
The majority declines to reach the issue of the prosecutor’s homophobic comments, concluding (1) that Neill’s counsel at trial was not constitutionally ineffective because he objected to the remarks and (2) without examining, or even quoting, the prosecutor’s remarks, that Neill’s counsel on direct appeal was not ineffective “[bjecause none of the[ ] prosecutorial misconduct claims have merit.” (Majority Op. at 1195 n. 5.) While the objection mitigates the ineffectiveness of Neill’s counsel’s performance at Neill’s trial,2 nothing limits review of that same lawyer’s effectiveness in failing to raise the issue during his subsequent representation of Neill on direct appeal.3
The Sixth Amendment requires that criminal defendants receive effective assistance of counsel both at trial and during a direct appeal as of right. See Evitts v. Lucey, 469 U.S. 387, 396, 105 S.Ct. 830, 83 L.Ed.2d 821 (1985) (“A first appeal as of right ... is not adjudicated in accord with due process of law if the appellant does not have the effective assistance of an attorney.”). Claims of ineffective assistance of appellate counsel are reviewed under the test in Strickland v. Washington, 466 U.S. 668, 687, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), and must demonstrate that “appellate counsel’s performance was deficient and that the deficient performance prejudiced [appellant’s] defense.” Smith v. Massey, 235 F.3d 1259, 1274 (10th Cir.2000). When the claim of appellate counsel’s ineffectiveness is based on the omission of an issue on appeal, our review is “highly deferential,” Moore v. Gibson, 195 F.3d 1152, 1180 (10th Cir.1999), cert. denied, 530 U.S. 1208, 120 S.Ct. 2206, 147 L.Ed.2d 239 (2000), and we cannot fault appellate counsel for failing to raise every conceivable issue on appeal, Cook, 45 F.3d at 394-95 (noting that appellate counsel properly weeds out weak issues on appeal so as not to detract a judge’s attention from stronger issues).
Nevertheless, appellate counsel renders ineffective assistance by “omitting a ‘dead-bang winner,’ even though counsel may have presented strong but unsuccessful claims on appeal.” Id. at 395 (citing Page, 884 F.2d at 302). A “dead-bang winner” “is an issue which was obvious from the trial record ... aoid one which would have resulted in a reversal on appeal.” Id. Subsequent cases have offered other formulations of the necessary showing to support a claim of ineffective assistance of counsel .based on omitting issues on appeal. See Johnson v. Gibson, 169 F.3d 1239, 1251 (10th Cir.1999) (whether the omitted claims were “clearly meritorious”); Jackson v. Shanks, 143 F.3d 1313, 1321 (10th Cir.1998) (whether the omitted claims were “obvious winners”); Banks v. Reynolds, 54 F.3d 1508, 1515 n. 13 (10th Cir.1995) (whether the issue “is obvious from the trial record and [is] one which probably would have resulted in a reversal on appeal” (emphasis added)). Whatever the exact nature of the “dead-bang winner” test, it requires a close look at the merits of the omitted claim. See, e.g., Hooks v. Ward, 184 F.3d 1206, 1221 (10th Cir.1999) (“When considering a claim of ineffective *1201assistance of appellate counsel for failure to raise an issue, we look to the merits of the omitted issue.”); Banks, 54 F.3d at 1515 (“When a habeas petitioner alleges that his counsel was ineffective for failing to raise an issue on appeal, we examine the merits of the omitted issue”). I fail to understand how the majority can summarily conclude that a claim of ineffective assistance of appellate counsel lacks merit absent even a cursory examination of the underlying basis for the claim. See Hawkins v. Hannigan, 185 F.3d 1146, 1152 (10th Cir.1999) (“When a habeas petitioner alleges that his appellate counsel rendered ineffective assistance by failing to raise an issue on direct appeal, we first examine the merits of the omitted issue.”); cf. Zant v. Stephens, 462 U.S. 862, 885, 103 S.Ct. 2733, 77 L.Ed.2d 235 (1983) (“[T]he severity of the [death] sentence mandates careful scrutiny in the review of any colorable claim of error.”).4
Turning, then, to the nature of the prosecutor’s comments, I think they are susceptible of only one possible interpretation: among other factors, Neill should be put to death because he is gay. The prosecutor urged the jury to consider “some things that ... depict the true person, what kind of person” Neill is. (V Trial Tr. at 1286.) According to the prosecutor, the “true person,” the “kind of person” Neill is can be summed up in four words: “He is a homosexual.” (Id.)
As the prosecutor knew, emphasizing that Neill was gay likely had a tremendous negative impact on jurors. See, e.g., Stockton v. Murray, 41 F.3d 920, 927 (4th Cir.1994) (the defendant’s “alleged homosexual activities ... had the potential to seriously prejudice the jury”); United States v. Ham, 998 F.2d 1247, 1252 (4th Cir.1993) (“We accept without need of extensive argument that implications of ... homosexuality ... unfairly prejudice a defendant.” (footnote omitted)); United States v. Gillespie, 852 F.2d 475, 479 (9th Cir.1988) (“Evidence of homosexuality is extremely prejudicial.”); Cohn v. Papke, 655 F.2d 191, 194 (9th Cir.1981) (introduction of evidence of homosexuality creates a “clear potential that the jury may have been unfairly influenced by whatever biases and stereotypes they might hold with regard to homosexuals”); Developments in the Law — Sexual Orientation and the Law: II. Gay Men and Lesbians and the Criminal Justice System, 102 Harv. L.Rev. 1519, 1552 (1989) (“[H]omophobia and anti-gay prejudice permeat[e] popular culture.”); cf. United States v. Bautista, 145 F.3d 1140, 1152 (10th Cir.1998) (“[T]estimony of [the victim’s] homosexuality was irrelevant and potentially highly prejudicial. We fail to discern how the victim’s sexual orientation is relevant to the charge of second-degree murder....”); Sexual Orientation Fairness Subcomm., Judicial Council of Cal., Sexual Orientation Fairness in California Courts 25-26 (2001), available at http ://www. courtinfo .ca.gov/programs/access/documents/report.pdf (reporting the results of a study finding that “56 percent *1202of gay and lesbian court users in a contact in which sexual orientation became an issue reported observing or experiencing a range of negative experiences.... Specifically, 36 percent heard negative comments about someone else; 29 percent heard negative remarks arising from a case; 23 percent heard negative comments about themselves; 26 percent experienced or heard ridicule, snickering, or jokes about lesbians and/or gay men; and 25 percent heard other negative remarks.”).
The State seeks to explain away the prosecutor’s comments by claiming that Neill’s “sexual orientation was relevant to the issues in the case because it was the problems arising from the relationship with Johnson that provided the motivation for the robbery and murders.” (Appellee’s Br. at 42.) It is true that Neill did not attempt to hide the nature of his romantic relationship and that he argued that problems in the relationship were a factor contributing to his criminal activity. However, the State’s contention suffers from a fundamental misunderstanding of the purpose of Neill’s mitigating evidence. Neill introduced evidence of the emotional and financial stresses in his relationship to show that the robbery was the act of a desperate man, not that of a depraved killer. The prosecutor was within his rights in criticizing the nature of Neill’s mitigating evidence, as when he stated “Any of you been in a relationship that broke up? Did that justify or warrant you going out and killing four people and shooting three others? ... The fact that you’re losing a lover does not put you in the emotional state where it would justify this.” (V Trial Tr. at 1283.) However, references to the fact that Neill was “a vowed [sic] homosexual” (id.) and had a “gay lover” (id.) add nothing to the critique and only serve to highlight the irrelevant and prejudicial fact of Neill’s sexual orientation. The prosecutor’s comments a few pages later in the transcript are totally unresponsive to Neill’s mitigation evidence and are designed solely to inflame the jury’s prejudices:
But just generic, just put in the back of your mind what if I was sitting in judgment on this person without relating it to Jay Neill, and I’d like to go through some of the things that to me depict the true person, what kind of person he is. He is a homosexual. The person you’re sitting in judgment on — -disregard Jay Neill. You’re deciding life or death on a person that’s a vowed [sic] homosexual.
(Id. at 1286.)
As an additional justification for the prosecutor’s comments, the State notes that it was Neill himself who “introduced the issue of his sexual orientation.” (Ap-pellee’s Br. at 42.) To my mind that argument is no different from claiming that a Jewish defendant opens the door to a prosecutor’s anti-Semitic arguments by wearing a yarmulke in the presence of jurors.
At their core, the prosecutor’s statements are tantamount to urging the jury to return the death sentence because a defendant fits within any other group that has been the target of prejudice and discrimination. See generally Debra T. Landis, Annotation, Prosecutor’s Appeal in Criminal Case to Racial, National, or Religious Prejudice as Ground for Mistrial, New Trial, Reversal, or Vacation of Sentence — Modem Cases, 70 A.L.R.4th 664, 1989 WL 571785 (1989) (discussing the issue and citing cases). Directing the jurors to focus on Neill’s homosexuality when deciding his punishment was an obviously improper appeal to their prejudices. See American Bar Association Standards for Criminal Justice 3-5.8(c) (3d ed. 1993) (“The prosecutor should not make arguments calculated to appeal to the prejudices of the jury.”). A civilized system of criminal justice can have no place for such actions. See, e.g., Gardner v. Florida, 430 *1203U.S. 349, 358, 97 S.Ct. 1197, 51 L.Ed.2d 393 (1977) (plurality) (“It is of vital importance to the defendant and to the community that any decision to impose the death sentence be, and appear to be, based on reason rather than caprice or emotion.”).
The prejudicial impact of bigoted comments at the penalty phase of a capital trial was aptly described by the Nevada Supreme Court. In Dawson v. State, 103 Nev. 76, 734 P.2d 221, 223 (1987), the prosecutor made arguably irrelevant and racist remarks during Dawson’s sentencing. The court vacated the defendant’s death sentence, observing that “[i]n a capital sentencing proceeding before a jury, the jury is called upon to make a ‘highly subjective, unique, individualized judgment regarding the punishment that a particular person deserves.’ ” Id. (quoting Turner v. Murray, 476 U.S. 28, 33-34, 106 S.Ct. 1683, 90 L.Ed.2d 27 (1986) (further internal quotation omitted)). This special role of the jury in the penalty phase and the irrevocable nature of a death sentence result in a different calculation of prejudice in capital cases. “Because of the delicate task which the trier of fact has in weighing the mitigating circumstances against the aggravating circumstances, the kind and level of prejudice which might not require reversal of a conviction may be sufficient to require reversal of a death penalty.” Id. at 224.
Finally, I note the ineffectiveness of Neill’s counsel on appeal for failing to raise the prosecutorial misconduct claim is highlighted by several additional facts: the prosecutor’s comments were “obvious from the trial record ... [and] ‘leaped out upon even a casual reading of [the] transcript,’ ” Cook, 45 F.3d at 395 (quoting Matire v. Wainwright, 811 F.2d 1430, 1438 (11th Cir.1987)); Neill’s counsel — who represented Neill both at trial and on appeal — was present in the courtroom when the comments were made; and Neill’s counsel knew that Neill’s homosexuality was potentially prejudicial, as exemplified by his asking some (but not all) members of the jury about their attitudes toward homosexuals during voir dire. (See, e.g., I Trial Tr. at 149-62, 190, 199, 210, 223; II Trial Tr. at 258, 268.)
I cannot sanction — because I have no confidence in — a proceeding tainted by a prosecutor’s request that jurors impose a death sentence based, even in part, on who the defendant is rather than what he has done. For the reasons described, I believe Neill’s appellate counsel omitted a “dead-bang winner” on appeal, and Neill has thus shown cause and prejudice excusing his procedural default.
II
My confidence in Neill’s sentence is further undermined by the likelihood that he was denied an impartial jury due to the failure of both the court and his trial counsel to inquire on voir dire whether three members of the jury were predisposed to impose the death penalty. Neill proffers affidavits prepared by the Oklahoma Indigent Defender System (“OIDS”) reporting statements made by two of the jurors, Rusella Loggins and Glen Nelson Hyde, III, during interviews with OIDS investigators. Loggins told an investigator that she believed death was the only appropriate punishment for murder under any circumstances. Hyde, the jury foreman, told another investigator that when a person takes a life, he or she deserves the same sentence. Neill also points to a third juror’s voir dire response in which he stated that he would “like” to impose a death sentence. (II Trial Tr. at 471.)
I agree with the majority that we are constrained to review Neill’s biased jury claim through the lens of ineffective assistance of counsel. Such claims are evaluated under the familiar two-pronged ap*1204proach of Strickland, 466 U.S. at 687, 104 S.Ct. 2052. Under that analysis, Neill must show his “counsel’s performance fell below an objective standard of reasonableness, and there is a reasonable probability that, but for counsel’s errors, the outcome of the proceedings would have been different.” Nguyen v. Reynolds, 131 F.3d 1340, 1347 (10th Cir.1997) (citing Kimmelman v. Morrison, 477 U.S. 365, 375, 106 S.Ct. 2574, 91 L.Ed.2d 305 (1986)). “[Counsel’s actions during voir dire are presumed to be matters of trial strategy.” Fox v. Ward, 200 F.3d 1286, 1295 (10th Cir.2000). Consequently, claims of ineffective assistance of counsel at voir dire cannot succeed “unless counsel’s decision is shown to be so ill chosen that it permeates the entire trial with obvious unfairness.” Nguyen, 131 F.3d at 1349.
Against that background is the requirement that persons the state wishes to execute be convicted and sentenced by impartial juries. In particular, the Supreme Court held in Morgan v. Illinois, 504 U.S. 719, 729, 112 S.Ct. 2222, 119 L.Ed.2d 492 (1992), that “a capital defendant may challenge for cause any prospective juror who [will automatically impose the death penalty], If even one such juror is empaneled and the death sentence is imposed, the State is disentitled to execute the sentence.” Further, the Court noted that although “[t]he Constitution ... does not dictate a catechism for voir dire, ... part of the guarantee of a defendant’s right to an impartial jury is an adequate voir dire to identify unqualified jurors.” Id. (citations omitted). We have echoed the Court’s concern about adequate voir dire in death penalty cases. United States v. Chanthadara, 230 F.3d 1237, 1269 (10th Cir.2000) (“[B]ecause the jurors are vested with greater discretion in capital cases, the examination of prospective jurors must be more careful than in non-capital cases.”); Hale v. Gibson, 227 F.3d 1298, 1318 (10th Cir.2000) (“[D]ue process requires a voir dire examination of a potential juror’s views on the death penalty.”). I read Morgan, Chanthadara, and Hale as not only permitting, at a party’s request, a thorough voir dire of potential jurors’ views of the death penalty, but actually requiring that such questioning take place as a matter of due process.
Neill’s counsel’s performance was deficient based on the requirement of an adequate voir dire. Despite receiving a response that indicated that Juror Han-nabass “would like to” sentence Neill to death, “[njeither defense counsel nor anyone else further questioned this juror concerning his ability to consider imposing a sentence less than death.” (Majority Op. at 1195.) This deficiency was exacerbated by the fact that the entire trial revolved around the penalty phase. Neill’s trial counsel basically conceded guilt (a reasonable decision considering the overwhelming evidence against Neill). That situation should have made Neill’s counsel especially vigilant concerning jurors’ attitudes about the death penalty as it was certain that they would be called upon to decide whether death was the appropriate punishment. Under those circumstances, failing to ask jurors their views on the death penalty could not have been a viable strategy.
In addition, I question the majority’s contention that “the record is insufficient to permit this court to determine whether defense counsel’s failure to ask only these three jurors whether they would automatically vote for a death sentence was strategic.” (Majority Op. at 1194.) To the contrary, the record in this case clearly indicates that it was counsel’s strategy to question jurors about their views on the death penalty. In letters to Neill, his counsel stated that he would “concentrate” on voir dire (Post-Conviction Application App. E at 4) and that he was aware *1205of Morgan and would “just ask enough [at voir dire] to preserve error for review” (id. at 14). In addition, someone, “either defense counsel or the prosecutor[,] asked all ... prospective jurors” other than the three who are the subjects of the proffered affidavits “whether they would consider imposing a sentence less than death.” (Majority Op. at 1194.) The unexplained failure to question these jurors — which flies in the face of everything in the record — can be viewed as nothing other than objectively deficient performance by Neill’s counsel.
As for the prejudice prong, the majority concludes that the proffered affidavits are both hearsay and too conclusory to merit habeas relief. While I agree that standing alone the affidavits do not merit vacating Neill’s sentence,5 when combined with the prosecutor’s homophobic remarks, they further undermine confidence in the jury’s verdict and lend additional support to my conclusion that Neill’s death sentence cannot stand.
Ill
I would grant habeas relief and vacate Neill’s sentence.
. I have serious reservations about the viability of the "dead-bang winner” standard in light of the Supreme Court’s holding in Smith v. Robbins, 528 U.S. 259, 285, 120 S.Ct. 746, 145 L.Ed.2d 756 (2000), that "the proper standard” for reviewing a claim of ineffective assistance of appellate counsel "is that enunciated in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984).” The Court required nothing more than the usual showing for an ineffective assistance claim: a habeas petitioner "must first show that his counsel was objectively unreasonable ... in failing to find arguable issues to appeal — that is, that counsel unreasonably failed to discover nonfrivolous issues” and "that, but for his counsel’s [deficient performance] ..., he would have prevailed on his appeal.” Smith, 528 U.S. at 285, 120 S.Ct. 746. Nevertheless, my analysis proceeds under the "dead-bang winner” standard because I believe Neill should prevail regardless.
. In addition to his objection, Neill’s counsel could have moved for a mistrial.
. The real significance of a successful ineffective assistance of appellate counsel claim is that it provides the necessary “cause” to excuse the procedural default barring habeas review of the underlying issue. See, e.g., Murray v. Carrier, 477 U.S. 478, 488, 106 S.Ct. 2639, 91 L.Ed.2d 397 (1986) ("Ineffective assistance of counsel ... is cause for a procedural default.”); Banks v. Reynolds, 54 F.3d 1508, 1514 (10th Cir.1995) ("A habeas petitioner may establish cause for his procedural default by showing that he received ineffective assistance of counsel in violation of the Sixth Amendment.”).
. We have recognized that reviewing the merits of a defendant's claim "in order to determine whether he received ineffective assistance from his appellate counsel” "places us in an awkward position.” Banks, 54 F.3d at 1516. As noted above, "[a]ssuming [Neill] demonstrates ineffective appellate assistance, his procedural default will be excused and we may then review the merits of his claims.” Id. (citations omitted). However, in considering whether Neill received ineffective assistance of appellate counsel, we end up considering the merits of his defaulted prosecutorial misconduct claim to determine whether it is a "dead-bang winner.” Id. at 1515. "Notwithstanding the apparent circularity of this review, our ultimate inquiry is central and straightforward: is our confidence in the outcome of [Neill’s] ... sentence undermined by the” alleged prosecutorial misconduct? Id. at 1516.
. Even considered in isolation, the affidavits do raise sufficiently serious allegations to require an evidentiary hearing. Neill “is entitled to an evidentiary hearing 'if his allegations, if true and not contravened by the record,' entitle him to habeas relief.” Walker v. Gibson, 228 F.3d 1217, 1231 (10th Cir.2000) (quoting Mayes v. Gibson, 210 F.3d 1284, 1287 (10th Cir.2000)). Neill meets that standard: the Supreme Court has held that "[i]f even one such [biased] juror is empaneled and the death sentence is imposed, the State is disentitled to execute the sentence.” Morgan, 504 U.S. at 729, 112 S.Ct. 2222 (emphasis added). In this case, the deficient performance of Neill’s trial counsel potentially permitted three biased jurors to sit on the jury that imposed his death sentence. An evidentiary hearing would address the majority's concerns with the shortcomings of Neill's affidavits. To avoid hearsay problems the jurors themselves could be called to testify. Full questioning would “provid[e] the context in which these jurors made the[] statements [in their affidavits]” and would supply "the time frame during which they ... held these beliefs.” (Majority Op. at 1195.) Such a hearing is not necessary, in my view, because Neill is entitled to resentencing outright based on his prosecutorial misconduct claim.