OPINION OF THE COURT
GREENBERG, Circuit Judge.I. INTRODUCTION
This matter is before the court on Charles J. Goldblum’s appeal from the denial of his second petition for a writ of habeas corpus. Goldblum, who is currently serving a life sentence following his conviction for first-degree murder, filed his second petition after receiving our authorization to do so under 28 U.S.C. § 2244(b)(3)(A). The district court, in adopting a magistrate judge’s Report and Recommendation, dismissed the second petition based on Goldblum’s failure to satisfy the requirements applicable to second petitions under 28 U.S.C. § 2244(b)(4) and its predecessor, the abuse-of-the-writ doctrine, which as we will discuss, was implicated because Goldblum filed his first habeas corpus application before the enactment of the Antiterrorism and Effective Death Penalty Act of 1996 (“AED-PA”).
Goldblum believes that the magistrate judge erred in three ways: she (1) was required to conduct an evidentiary hearing to determine whether he abused the writ; (2) applied the wrong legal standard under the “cause” element of the abuse-of-the-writ doctrine; and (3) wrongly found that Goldblum is not “actually innocent” of the murder for which he has been convicted, which actual innocence would have excused his noncomplianee with the procedures ordinarily required for habeas corpus proceedings to avoid his petition being barred by the abuse-of-the-writ doctrine. We find *206Goldblum’s arguments unpersuasive and thus will affirm the order of the district court, thereby upholding the dismissal of his second habeas application.
II. FACTS AND PROCEDURAL HISTORY
The events and procedural history leading up to this appeal are quite complicated and lengthy. We therefore will discuss them only as they relate to this appeal.
A. The Underlying Convictions
At Goldblum’s Pennsylvania state-court trial on charges of murder, arson, and insurance fraud, the prosecution put forth the following facts. In 1974, Clarence Miller met George Wilhelm. At that time, they discussed Wilhelm’s interest in purchasing land in North Carolina on which Wilhelm planned to search for semiprecious stones and gemstones. Miller discovered that the land that Wilhelm was interested in purchasing was federal forest land and was not for sale. Miller, however, told Wilhelm that he would use his “political connections” to assist him to purchase the land. Miller, however, did not have these so-called “political connections.” Instead, he devised a scheme to defraud Wilhelm in which he planned to tell Wilhelm that he would work out a special political deal with the help of a United States senator to obtain the government-owned land.1
Miller contacted Thaddeus Dedo and Goldblum to assist him in executing this fraud. Dedo, under Goldblum’s guidance, impersonated an actual member of the senator’s staff, Ken Manella, and made several phone calls to Wilhelm confirming the deal leading Wilhelm to give Miller a series of payments totaling approximately $20,000 for the consideration to put through the purchase. In exchange, Miller gave Wilhelm fake deeds to land in North Carolina that Goldblum apparently drafted.
The scheme began to unravel when Wilhelm went to the senator’s office to meet with “Ken Manella.” Wilhelm immediately became suspicious that he had been defrauded when the real Ken Manella appeared. Wilhelm reported his suspicions to the Federal Bureau of Investigation. The FBI, however, terminated its investigation when Goldblum and Miller persuaded Wilhelm to withdraw his complaint in exchange for Goldblum’s and Miller’s promise that Wilhelm would get his money back. Wilhelm did withdraw it by asserting to the FBI that his complaint was a hoax.
The plot then thickened as the money to repay Wilhelm was not readily available. In order to raise the money, Wilhelm agreed with Goldblum to participate in an insurance fraud scheme in which Wilhelm would set fire to a restaurant Goldblum leased and operated, but Goldblum’s parents owned. In return, Goldblum was to pay Wilhelm $3,500 in addition to the money taken from him in the land fraud. The restaurant burned to the ground as a result of arson on November 30, 1975. Gold-blum paid Wilhelm $100, but when no one paid Wilhelm the remaining money, Wilhelm, who surely was not short of nerve, began pressing Goldblum for payment and threatening him that he would go to the authorities.
Goldblum, understandably in Anew of Wilhelm’s previous contact Avith the FBI, obtdously took Wilhelm’s threats seriously for on February 8, 1976, he told Miller *207that he intended to beat Wilhelm up to discourage him from pressing him for the debt payment or going to the authorities. Miller agreed to assist in this plan by luring Wilhelm to the top floor of a parking garage in exchange for $50 and Miller did so by telling Wilhelm that Goldblum had the money he was due. On February 9, 1976, the three of them, with Wilhelm driving, Miller sitting in the front passenger seat, and Goldblum sitting in the back seat behind the driver, drove to the top floor of a parking garage in downtown Pittsburgh.
This case revolves around what happened next, a matter in some dispute. Miller contended at Goldblum’s trial that Goldblum struck Wilhelm in the back of the head with a wrench and Wilhelm fell out the car, at which time Goldblum began stabbing him with a grass shear blade and Wilhelm fell over a wall. Goldblum, on the other hand, claims that Miller and Wilhelm got into a fight while in the car, leading to the stabbing, following which Wilhelm opened his door to the car and fell to the ground, at which time Miller flipped him over the wall. In any event, Goldblum and Miller left the scene together and agreed that they would say that they only had seen Wilhelm earlier in the evening, but they were not with him at the time of the murder.
Wilhelm was found later that night, crying for help. When police arrived at the scene, Wilhelm said to them, “Clarence Miller did this to me.” Wilhelm died a few hours later. The police arrested Miller and when he, in turn, implicated Goldblum, police arrested Goldblum as well. Gold-blum was released on bail. The police, however, fortunately engaged in a surveillance of Goldblum, during which he was observed arranging for Miller’s murder with an undercover detective. Consequently, they arrested Goldblum on a charge for that new offense and returned him to jail.
The authorities filed a complaint against Goldblum in the Court of Common Pleas of Allegheny County charging him with murder and voluntary manslaughter of Wilhelm, criminal conspiracy in relation to the fraudulent land deal, and arson and criminal solicitation to commit arson of the restaurant.2 They also charged Miller in the stabbing death of Wilhelm.
Goldblum proceeded to trial. The prosecution argued that Goldblum killed Wilhelm with the motive to silence him regarding the arson that they had committed. Goldblum, on the other hand, argued that he was not guilty of anything.3 The defense focused on Miller’s obviously suspect credibility, as he was the prosecution’s central witness, as well as the physical evidence that, according to Gold-blum, tended to establish that Miller did the stabbing. In particular, Goldblum’s attorney argued to the jury that while blood was found on Miller’s overcoat, no blood was found on Goldblum’s clothes. He also introduced circumstantial evidence relating to the pattern of the blood *208spatter on the dashboard of the car that the three men had occupied, which suggested that the person sitting to the right of Wilhelm, i.e., Miller, not the person in the back seat, i.e., Goldblum, did the stabbing. His attorney also introduced into evidence Wilhelm’s dying declaration that “Clarence Miller did this to me.”
The court instructed the jury both on the theory that Goldblum was guilty of murder in the first degree for the direct assault and the theory that he was an accomplice to murder in the first degree. On August 30,1977, a jury found Goldblum guilty of murder in the first degree for the death of Wilhelm, as well as conspiracy to commit theft by deception, arson, and criminal solicitation to commit arson.4 The court did not ask the jury to determine whether Goldblum directly participated in the assault or was an accomplice to Miller’s action, and the jury did not make a finding on this point. The court sentenced Goldblum to life in prison on the murder conviction and an additional 15 to 30 years imprisonment for the other offenses.
Goldblum was unsuccessful on his direct appeal during which he argued that Miller suffered from organic brain damage that impacted on his ability to distinguish fact from fiction. See Commonwealth v. Gold-blum, 287 Pa.Super. 544, 427 A.2d 258 (1980) (Superior Court affirming convictions except conspiracy count), rev’d in part, 498 Pa. 455, 447 A.2d 234 (1982) (Supreme Court reversing Superior Court on conspiracy count and affirming other convictions). The state courts similarly denied his petition for post-conviction relief.
B. Goldblum’s First Federal Habeas Corpus Petition
Goldblum subsequently sought relief in the federal courts, filing his first petition for a writ of habeas corpus in the district court on July 14,1989. In that petition, he presented the following two claims:
1. Whether the denial of [Goldblum’s] pretrial application for psychiatric examination of the prosecution’s only eyewitness [Miller], coupled with the denial of his motion for [a] new trial based on after discovered evidence providing a basis for attacking the credibility of that witness, together denied [Goldblum] due process under the Fourteenth Amendment.
2. Whether the admission into evidence of the out-of-court declarations of Wilhelm that [Goldblum] had participated in the land fraud and that Wilhelm had participated in the arson of Goldblum’s restaurant deprived him of his right to confront the witnesses against him guaranteed by the Sixth and Fourteenth Amendments to the United States Constitution.
The district court denied his petition on the merits, following which on Goldblum’s appeal we affirmed the order of the district court without a published opinion on November 26, 1991. Goldblum v. Fulcomer, 950 F.2d 722 (3d Cir.1991) (table). The Supreme Court denied Goldblum’s application for a writ of certiorari on April 27, 1992. Goldblum v. Fulcomer, 503 U.S. 1005, 112 S.Ct. 1760, 118 L.Ed.2d 423 (1992).
C. Back to the State Courts
After the completion of the unsuccessful habeas corpus proceedings, Goldblum returned to the state courts, this time filing a petition for post-conviction relief under the *209Post-Conviction Relief Act (“PCRA”) on January 12, 1996, wherein he raised a litany of issues, including the discovery of evidence relating to forensic proof of blood spatter, Miller’s alleged post-trial confessions, and ineffective assistance of counsel relating to the failure of his trial attorney to investigate evidence pertaining to the blood spatter and present the expert testimony of Dr. Cyril Wecht,5 a forensic pathologist, who would have testified that based on the physical evidence Goldblum could not have been the assailant. In support of this physical evidence theory, Gold-blum filed Dr. Wecht’s affidavit in which he “concluded to a reasonable degree of medical certainty that Mr. Goldblum was not the individual who inflicted the fatal stab wounds to Mr. Wilhelm.” App. at 21. Dr. Wecht based this conclusion substantially on the location of the blood spatter on the dashboard, as well as the lack of blood on Goldblum’s clothes. In addition, Goldblum submitted two other expert witnesses’ affidavits expressing opinions similar to those of Dr. Wecht. Goldblum subsequently added a claim to his petition that his trial attorney was ineffective for failing to object to the trial court’s instruction on accomplice liability.
On February 12, 1997, the state court dismissed the petition without holding an evidentiary hearing, finding that all of the claims Goldblum raised either had been litigated previously or were too old to be considered. On Goldblum’s appeal, the Superior Court affirmed, concluding that all his claims were waived, previously litigated, or meritless, except the claim that his trial attorney was ineffective for failing to call Dr. Wecht as a witness at the trial. The Superior Court remanded the case to the trial court for an evidentiary hearing solely on that issue.
The trial court held an evidentiary hearing on October 18, October 19, and December 19, 2000. The first witness was Gold-blum’s trial attorney, H. David Rothman. Rothman testified that he did not conduct an investigation with respect to the blood spatter evidence because “the police did not photograph or preserve the blood spatters that were found on the dashboard of Mr. Wilhelm’s car.” He had not consulted a pathologist before the trial because he believed, based on the literature he had seen, that the evidence that could be developed would have been unreliable. Therefore, Rothman argued at trial, without the aid of expert testimony, that based on the circumstantial physical evidence showing that the blood spatter went in the direction of Miller, Miller was the killer.
Dr. Wecht testified next, and on direct examination, he expounded upon his affidavit. On cross-examination, however, he conceded that there were other possible factual scenarios that would explain the pattern of the blood spattering.6 The court refused to take testimony from the other forensic experts who would have supported Dr. Wecht’s opinion, and also would not take the testimony of an expert in police investigation who was critical of the lack of photographs of the dashboard and an attorney who would have testified as to the actions Goldblum’s trial attorney should have taken to investigate this case.
The Commonwealth presented a rebuttal witness, Toby Wolson, a forensic biologist employed by the Miami-Dade Florida Police Department, who was testifying independently from that relationship as a *210forensic consultant. Wolson testified that the limited description of the blood spatter prevented both Dr. Wecht and him from reaching a reliable conclusion as to the identity of the assailant. Detective Ron Freeman had been the only witness at the trial who explained the location of the blood spatter. Freeman testified at the trial that:
When I saw there were a small line of blood droplets, and it was not a lot of blood there, but they were discernable droplets and they started on the — toward the driver’s side was the largest spot, and then they descended into smaller circles, and each circle has what is called ‘a tail’ and the ‘tail’ was facing toward the passenger side of the automobile and that indicated that blood came from left to right, as I faced it or traveled from the driver’s side of the automobile to the passenger’s side of the automobile.
On August 22, 2001, the state court issued an opinion and order denying Gold-blum PCRA relief, holding:
[T]his Court is mindful that no photographs or other evidence of the blood stain existed. The testimony of both Dr. Wecht and Mr. Wolson was premised on the fact that Detective Freeman’s brief description [at the trial] of the blood stain was accurate. A review of the testimony presented indicated that both experts were hesitant to declare their findings absolute without being able to see the blood stain in question. Although the experts did make tentative findings, their testimony essentially amounted to speculation due to their inability to make conclusive findings.
App. at 265-66. The court explained its decision to exclude the other potential witnesses on the basis that the limited scope of the remand did not give it authority to hear that testimony.
Goldblum appealed, but on October 24, 2002, the Superior Court affirmed the denial of PCRA relief, concluding that due to the “fundamentally inconclusive” nature of Dr. Wecht’s testimony, the court could not conclude that the outcome of the trial would have been different had Dr. Wecht testified at the trial. Id. at 280. The Superior Court also found that the PCRA court’s decision to limit the testimony based on the scope of the remand was proper. Subsequent petitions for an allowance of appeal in the Supreme Court of Pennsylvania, Commonwealth v. Goldblum, 573 Pa. 689, 825 A.2d 637 (2003), and for a writ of certiorari with the United States Supreme Court, Goldblum v. Pennsylvania, 540 U.S. 1119, 124 S.Ct. 1067, 157 L.Ed.2d 914 (2004), were denied.
D. Goldblum’s Second Federal Habeas Corpus Petition and Appeal
On February 26, 2004, Goldblum filed a motion with us seeking authorization under 28 U.S.C. § 2244(b)(3) to file a second petition for a writ of habeas corpus, which we granted on March 29, 2004. Goldblum promptly filed his second application for a writ of habeas corpus in the district court on April 2, 2004, asserting the following claims as recast before the magistrate judge:
1. Trial counsel (as well as successor state counsel) were ineffective for failing to investigate, preserve and produce vital scientific evidence of blood spatter that would have proven that the Commonwealth’s principal witness, Clarence Miller, was the person who stabbed and killed the victim, Mr. Wilhelm.
2. Trial counsel (as well as successor state counsel) was ineffective for failing to object to the state trial court’s erroneous and prejudicial instruction regarding accomplice lia*211bility. Specifically, there was no objection to (a) the trial court’s failure to instruct the jury that it could not find [Goldblum] as an accomplice unless they found beyond a reasonable doubt that [Goldblum] acted with the specific intent to kill in acting as an accomplice or (b) the trial court’s instruction to the jury that Clarence Miller was an accomplice of [Gold-blum].
3. The trial court’s instruction to the jury on accomplice liability was constitutionally flawed and deprived [Goldblum] of due process of law.
4. The Commonwealth’s loss and/or destruction of the investigative files was done intentionally and with the purpose of depriving [Goldblum] of evidence that would support his legal claims, including his claim of innocence, all in violation of [Gold-blum’s] right to due process of law.
5. Newly discovered evidence regarding the Commonwealth’s principal witness, Clarence Miller, provides strong proof that [Goldblum] is innocent of the crime of murder and that the conviction was based on perjured testimony. Mr. Miller has admitted stabbing the victim and, while he continues to insist that [Goldblum] was also involved, his admissions to a Warden and the State Attorney General are entirely inconsistent with his trial testimony, and are supportive of [Goldblum’s] claims of innocence. In these circumstances, this evidence provides grounds for a new trial on grounds of due process of law.
6.The state courts failed to provide [Goldblum] with a full and fair post-conviction hearing on these claims and thereby denied him due process of law.
Appellant’s br. at 29.
On October 28, 2005, a magistrate judge to whom the district court assigned the matter, without holding an evidentiary hearing, issued a Report and Recommendation dismissing Goldblum’s second application for a writ of habeas corpus on the procedural ground that it did not satisfy the requirements for second petitions. Based on our opinion in In re Minarik, 166 F.3d 591 (3d Cir.1999), in which we discussed the retroactivity of the AEDPA’s gatekeeping provision, the magistrate judge examined Goldblum’s second application under both section 2244 of the AEDPA and its predecessor, the abuse-of-the-writ doctrine. She began with her analysis under the AEDPA standard found in 28 U.S.C. § 2244(b),7 and recommended dismissal of Goldblum’s six claims because Goldblum either had presented them in his first habeas corpus application, thus re*212quiring their dismissal under section 2244(b)(1), or he had not presented them, and the claims did not rely on a new rule of constitutional law or meet the two-part standard of section 2244(b)(2)(B).
The magistrate judge next analyzed the claims under the pre-AEDPA standard “to determine if there [was] a conflict” between pre- and post-ADEA standards in which event the claims, if any, the pre-AEDPA regime did not bar would have to be addressed on their merits in order to circumvent “a retroactivity issue” with respect to the application of the AEDPA. We discuss this retroactivity problem in detail below. The magistrate judge set forth the abuse-of-the-writ standard discussed in Minarik, 166 F.3d 591, in which we concluded that there is an abuse of the writ precluding claims presented in second petitions unless (a) petitioner establishes “cause” for not including the claim in the first petition and “prejudice,” or (b) there would be a “fundamental miscarriage of justice” if the claim is not reviewed on its merits. She held that Goldblum did not establish “cause” for his failure to present his first claim (ineffective assistance of counsel relating to the blood spatter evidence) in the original habeas corpus petition, nor was there a fundamental miscarriage of justice as there was no physical evidence depicting the blood spatter which would have allowed counsel to pursue the defense. She gave a presumption of correctness to the PCRA court’s factual findings, rebuttable only by clear and convincing evidence under 28 U.S.C. § 2254(e)(1), with respect to the state court’s conclusion as to the speculativeness of Dr. Wecht’s testimony.
The magistrate judge also refused to grant Goldblum an evidentiary hearing under 28 U.S.C. § 2254(e)(2), as she found that the state court had developed the factual basis of his claims sufficiently in conducting a lengthy hearing on the issue of ineffective assistance and the state-court record included extensive exhibits. She further concluded that even if Dr. Wecht’s testimony had been introduced at trial, it certainly was not “outcome determinative” in light of the overwhelming evidence of guilt implicating Goldblum in the murder, and thus, there was no miscarriage of justice.
The magistrate judge concluded with respect to the second and third claims (ineffective assistance of counsel relating to his failure to object to the jury instruction regarding accomplice liability), that there was no “cause” as the law governing the instruction was known at the time that Goldblum filed his original habeas corpus petition. Nor was there a “miscarriage of justice” in which the alleged error “probably resulted in the conviction of one who is actually innocent” as the jury charge sufficiently informed the jury that specific intent was required to convict for first-degree murder on accomplice liability.
Similarly, as to the fifth claim (Miller’s admissions), the magistrate judge found that there was no “actual prejudice” as Miller’s out-of-court declaration, which we will describe below, did not exonerate Goldblum. In fact, Miller continues to assert that Goldblum participated in the killing, although Miller now admits that he, too, inflicted some of the wounds. The magistrate judge also recommended dismissal of Goldblum’s fourth and sixth claims for similar reasons, but we need not elaborate on this reasoning as this appeal does not focus on these claims.
Based on her conclusions that neither the AEDPA nor the pre-AEDPA abuse-of-the-writ doctrine permitted Goldblum’s second application, the magistrate judge recommended that the court dismiss Gold-blum’s second petition and that a certificate of appealability (“COA”) not be is*213sued. On December 13, 2005, the district court issued an order adopting the magistrate judge’s Report and Recommendation as the opinion of the court, dismissing the petition for a writ of habeas corpus, and denying the COA.
Goldblum appealed on January 12, 2006. On November 6, 2006, we issued a COA limited to the following question:
[Wjhether the District Court erred in concluding that Goldblum’s habeas petition constitutes an abuse of the writ, as Goldblum has shown that reasonable jurists would debate not only whether the District Court was correct in that procedural ruling but also whether his petition states a valid constitutional claim. See Slack v. McDaniel, 529 U.S. 473, 484, 120 S.Ct. 1595, 1604, 146 L.Ed.2d 542 (2000).
App. at 3.
Goldblum’s central arguments on this appeal are that the magistrate judge, and thus the district court, erred in the following three ways: she (1) was required to conduct an evidentiary hearing to determine whether Goldblum abused the writ; (2) applied the wrong legal standard under the “cause” element of the abuse-of-the-writ doctrine; and (3) wrongly found that Goldblum is not actually innocent of the murder for which he has been convicted, thus excusing his procedural noncompliance under the abuse-of-the-writ doctrine.
III. JURISDICTION AND STANDARD OF REVIEW
The district court had jurisdiction pursuant to 28 U.S.C. §§ 2241 and 2254(a). Upon issuing a limited COA on November 6, 2006, “on the question whether the District Court erred in concluding that Gold-blum’s habeas petition constitutes an abuse of the writ,” we have jurisdiction over the appeal of that specific issue pursuant to 28 U.S.C. §§ 1291 and 2253.
It is appropriate at this time to comment on the scope of the COA that we issued. The district court, by adopting the magistrate judge’s Report and Recommendation as the opinion of the court, dismissed Gold-blum’s second habeas application on the procedural ground that the AEDPA and its predecessor, the abuse-of-the-writ doctrine, barred his petition. It did not reach the merits of Goldblum’s constitutional claims, nor did it have the authority to do so until it first determined whether Gold-blum’s application satisfied section 2244’s requirements. See Benchoff v. Colleran, 404 F.3d 812, 816 (3d Cir.2005). At times, the magistrate judge was compelled to address some issues relating to the merits of Goldblum’s claims inasmuch as they were implicated in the determination of whether Goldblum’s claims met the threshold AED-PA and abuse-of-the-writ second petition standards, particularly his claims of actual innocence. This inquiry, however, does not take from our conclusion that the district court, by adopting the Report and Recommendation, dismissed the second petition on procedural grounds.
Thus, we reiterate that on November 6, 2006, we issued a COA:
[Ojn the question whether the District Court erred in concluding that Gold-blum’s habeas petition constitutes an abuse of the writ, as Goldblum has shown that reasonable jurists would debate not only whether the District Court was correct in that procedural ruling but also whether his petition states a valid constitutional claim. See Slack v. McDaniel, 529 U.S. 473, 484, 120 S.Ct. 1595, 1604, 146 L.Ed.2d 542 (2000).
It was certainly within our discretion to grant the limited COA under 28 U.S.C. § 2253(c)(3), inasmuch as the district court in the first instance should make a merits *214analysis of a habeas corpus petition if one is to be made.
Goldblum, nevertheless, contends in his brief that we “granted the COA on the abuse of the writ question as well as the validity of the constitutional claims.” Appellant’s br. at 7 n. 2. He urges that “no remand is necessary as the evidence of ineffectiveness of counsel cannot be disputed.” Id. at 29.
Goldblum misunderstands our COA. It is clear that we issued our COA only “on the question whether the District Court erred in concluding that Goldblum’s habe-as petition constitutes an abuse of the writ.” Though we also stated that the question was whether “Goldblum has shown that reasonable jurists would debate not only whether the District Court was correct in that procedural ruling but also whether his petition states a valid constitutional claim,” that language in no way suggests that we granted a COA on the merits of the constitutional issues. Rather, Slack v. McDaniel, 529 U.S. 473, 484, 120 S.Ct. 1595, 1604, 146 L.Ed.2d 542 (2000), required us to make that finding as it held:
When the district court denies a habeas petition on procedural grounds without reaching the prisoner’s underlying constitutional claim, a COA should issue when the prisoner shows, at least, that jurists of reason would find it debatable whether the petition states a valid claim of the denial of a constitutional right and that jurists of reason would find it debatable whether the district court was correct in its procedural ruling.
The purpose of this rule is to effectuate judicial efficiency as it would be a waste of the courts’ and the litigants’ resources to grant a COA on a procedural issue without a preliminary review of the underlying claims, for if the claims are obviously without merit and dismissal of the petition inevitably would be the result even if the petitioner overcame the procedural hurdles facing him, then further review of the procedural issue would be pointless. For this reason, we must make a preliminary review of the underlying claims even if we are granting a COA on the procedural issue only. But certainly in such situations we are not granting review of the merits of the petition, and should we find in favor of the applicant on the procedural issue, we would remand the matter to the district court to address the merits of the case.
Thus, in this case where we granted a COA only on the issue of whether Gold-blum abused the writ and because “[w]e may not consider issues on appeal that are not within the scope of the [COA],” we will not consider the merits of the underlying constitutional claims. See Villot v. Varner, 373 F.3d 327, 337 n. 13 (3d Cir.2004); see also 3d Cir. L.A.R. 22.1(b). If we were to find in favor of Goldblum on the procedural issue, we would remand the ease to the district court to consider the merits of the substantive claims. Of course, to the extent that we must take a preliminary look at the merits of the claims in the context of the abuse-of-the-writ determination, we will do so. See In re Williams, 330 F.3d 277, 282 (4th Cir.2003) (“While this determination [under section 2244] may entail a cursory glance at the merits-for example, an applicant cannot show that he would not have been convicted ‘but for constitutional error’ without adequately alleging some constitutional violation-the focus of the inquiry must always remain on the § 2244(b)(2) standards.”).
We review district court rulings on the abuse-of-the-writ doctrine de novo. See Zayas v. INS, 311 F.3d 247, 252 (3d Cir.2002). We review the district court’s decision to deny an evidentiary hearing for abuse of discretion. See Schriro v. Lan-*215drigan, — U.S. -, 127 S.Ct. 1933, 1939, 167 L.Ed.2d 836 (2007).
IV. DISCUSSION
A. Legal Framework of Second Applications Under the Abuse-of-the-Writ Standard and the AEDPA
Prior to the enactment of the AEDPA in 1996, the “doctrine of abuse of the writ define[d] the circumstances in which federal courts decline to entertain a claim presented for the first time in a second or successive petition for a writ of habeas corpus.” McCleskey v. Zant, 499 U.S. 467, 470, 111 S.Ct. 1454, 1457, 113 L.Ed.2d 517 (1991). When Goldblum filed his first federal habeas petition in 1991, the law encompassed the abuse-of-the-writ doctrine, which provided that a petitioner could prosecute another such petition only if he could “(1) show cause for, and prejudice from, the omission of his new claim or claims from his earlier petition (i.e., that his proceeding would not constitute an ‘abuse of the writ’), or (2) demonstrate ‘actual innocence.’ ” Minarik, 166 F.3d at 600.
The Supreme Court in McCleskey discussed the meaning of “cause” and “prejudice” prongs, and the “narrow exception” of “actual innocence,” as well as the burden-shifting framework when the abuse-of-the-writ doctrine is raised.
When a prisoner files a second or subsequent application, the government bears the burden of pleading abuse of the writ. The government satisfies this burden if, with clarity and particularity, it notes petitioner’s prior writ history, identifies the claims that appear for the first time, and alleges that petitioner has abused the writ. The burden to disprove abuse then becomes petitioner’s. To excuse his failure to raise the claim earlier, he must show cause for failing to raise it and prejudice therefrom as those concepts have been defined in our procedural default decisions. The petitioner’s opportunity to meet the burden of cause and prejudice will not include an eviden-tiary hearing if the district court determines as a matter of law that petitioner cannot satisfy the standard. If petitioner cannot show cause, the failure to raise the claim in an earlier petition may nonetheless be excused if he or she can show that a fundamental miscarriage of justice would result from a failure to entertain the claim.
McCleskey, 499 U.S. at 494-95, 111 S.Ct. at 1470 (internal quotation marks and citations omitted).
“Cause” requires “a showing of some external impediment preventing counsel from constructing or raising the claim.” Id. at 497, 111 S.Ct. at 1472 (internal quotation marks and citation omitted). “[T]he question is whether petitioner possessed, or by reasonable means could have obtained, a sufficient basis to allege a claim in the first petition and pursue the matter through the habeas process.” Id. at 498, 111 S.Ct. at 1472. Accordingly, “[i]f what petitioner knows or could discover upon reasonable investigation supports a claim for relief in a federal habeas petition, what he does not know is irrelevant. Omission of the claim will not be excused merely because evidence discovered later might also have supported or strengthened the claim.” Id.
“Once the petitioner has established cause, he must show actual prejudice resulting from the errors of which he complains.” Id. at 494, 111 S.Ct. at 1470 (internal quotation marks and citation omitted). “Actual prejudice” means “not merely that the errors at [ ] trial created a possibility of prejudice, but that they worked to his actual and substantial disadvantage, infecting his entire trial with *216error of constitutional dimensions.” United States v. Frady, 456 U.S. 152, 170, 102 S.Ct. 1584, 1596, 71 L.Ed.2d 816 (1982). See also Fischetti v. Johnson, 384 F.3d 140, 155 (3d Cir.2004).
Even if a petitioner cannot show cause and prejudice, “the failure to raise the claim in an earlier petition may nonetheless be excused if he or she can show that a fundamental miscarriage of justice would result from a failure to entertain the claim.” McCleskey, 499 U.S. at 494-95, 111 S.Ct. at 1470. A court, however, should exercise this authority only in a “narrow class of cases,” i.e., in “extraordinary instances when a constitutional violation probably has caused the conviction of one innocent of the crime.” Id. at 494, 111 S.Ct. at 1470. “To establish the requisite probability, the petitioner must show that it is more likely than not that no reasonable juror would have convicted him in light of the new evidence.” Schlup v. Delo, 513 U.S. 298, 327, 115 S.Ct. 851, 867, 130 L.Ed.2d 808 (1995).
In 1996, when the AEDPA became effective Congress changed the rules governing second petitions. The AEDPA instituted a “gatekeeping mechanism” which imposed strict procedural requirements and significantly altered the substantive showing an applicant had to make in order to proceed on new claims in a second petition.8 In Benchoff, 404 F.3d 812, we set forth the new procedural and substantive requirements which govern second or successive habeas petitions under the AEDPA. As a procedural matter, section 2244(b)(3)(A) requires a prospective applicant, before he may file a second or successive application in the district court, to “move in the appropriate court of appeals for an order authorizing the district court to consider the application.” A three-judge panel of the court of appeals may grant the motion authorizing the district court to consider the application “only if it determines that the application makes a prima facie showing that the application satisfies the requirements of this subsection.” 28 U.S.C. § 2244(b)(3)(B), (C). The court of appeals should make its determination no later than 30 days after the filing of the motion, id. at § 2244(b)(3)(D), and the decision “shall not be appealable and shall not be the subject of a petition for rehearing or for a writ of certiorari.” Id. at § 2244(b)(3)(E).
Section 2244(b)(2) requires that the court of appeals deny a motion to file a second or successive habeas petition unless:
(A) the applicant shows that the claim relies on a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court, that was previously unavailable; or
(B)(i) the factual predicate for the claim could not have been discovered previously through the exercise of due diligence; and
(ii) the facts underlying the claim, if proven and viewed in light of the evi*217dence as a whole, would be sufficient to establish by clear and convincing evidence that, but for constitutional error, no reasonable factfinder would have found the applicant guilty of the underlying offense.
Id. at § 2244(b)(2).
If the court of appeals determines that the procedural requirements for a second or successive petition have been met, the applicant then is permitted to file the second or successive application with the district court. But section 2244(b)(4) makes it clear that before a district court may consider the merits of the application, the petition must satisfy the substantive requirements for it. As we put it in Minarik, 166 F.3d at 600, “[t]hese substantive gatekeeping provisions were intended to reduce the universe of cases in which a habeas petition may go forward on a second or successive petition.” In this regard, a district court “shall dismiss any claim presented in a second or successive application that the court of appeals has authorized to be filed unless the applicant shows that the claim satisfies the requirements of [section 2244(b)(2) ].” 28 U.S.C. § 2244(b)(4). We have made it clear that “[u]nless both the procedural and substantive requirements of § 2244 are met, the District Court lacks authority to consider the merits of the petition.” Benchoff, 404 F.3d at 816.
In Minarik, 166 F.3d at 600, we were asked to determine whether section 2244’s gatekeeping provisions have an “impermissible retroactive effect” in cases in which the applicant filed his first petition prior to the AEDPA’s enactment and he filed his second petition after the AEDPA’s enactment. Guided by the Supreme Court opinions in Landgraf v. USI Film Products, 511 U.S. 244, 114 S.Ct. 1483, 128 L.Ed.2d 229 (1994), and Lindh v. Murphy, 521 U.S. 320, 117 S.Ct. 2059, 138 L.Ed.2d 481 (1997), we concluded that, “[i]n those cases where a prisoner in state custody had a right to prosecute a second or successive petition prior to AEDPA’s passage, but would be deprived of that right by these new gatekeeping provisions, ... applying the AEDPA standard would have a ‘genuine retroactive effect’ because it would attach a new and adverse consequence to pre-AEDPA conduct-the prosecution of the original proceeding.” Minarik, 166 F.3d at 600.
We distinguished AEDPA’s new procedural requirements from its substantive requirements. We decided that the procedural requirement-that an applicant first must seek authorization from a court of appeals before filing a second petition under 28 U.S.C. § 2244(b)(3)(A)-applied retroactively as it was a rule of procedure that did not “attach new legal consequences to events completed before its enactment.” Minarik, 166 F.3d at 599-600 (quoting Lindh, 521 U.S. at 320, 117 S.Ct. at 2059). However, with respect to the new substantive requirements, we found that if an applicant,
can show that he would have been entitled to pursue his second petition under pre-AEDPA law, then the Landgraf default rule prohibits applying AEDPA’s new substantive gatekeeping provisions to bar his claims. In the absence of such a showing, however, applying those standards to [an applicant] results in no genuine retroactive effect, and the AED-PA standard must be applied under the Supreme Court’s holding in Lindh that AEDPA’s habeas corpus amendments apply generally to cases filed after its effective date.
Id. at 602. Therefore we concluded that
anyone seeking to file a second or successive petition under 28 U.S.C. § 2254 after April 24, 1996, must move in the appropriate Court of Appeals for an or*218der authorizing the District Court to consider the application. When such a motion is filed by a petitioner whose previous petition was filed before that date, the Court of Appeals must apply the substantive gatekeeping standards of 28 U.S.C. § 2244(b) as amended by AEDPA unless such application would bar a second or successive petition that could have been considered by the District Court under the law existing at the time the previous petition was filed.
Id. at 609.
Thus, in this case, there is no question that the AEDPA’s procedural gatekeeping requirements for a court of appeals’ authorization apply and have been satisfied. We apply the substantive gatekeeping standard of 28 U.S.C. § 2244(b), “unless such application would bar a second or successive petition that could have been considered by the District Court under the law existing at the time the previous petition was filed.” Minarik, 166 F.3d at 609. Here, the district court by adopting the Report and Recommendation as the opinion of the court, concluded that both the substantive requirements of the AEDPA and the pre-AEDPA abuse-of-the-writ doctrine bar Goldblum’s claims. But, as we previously discussed, we granted a COA only with respect to the abuse-of-the-writ inquiry. Thus, under Minarik, should we agree with the district court that Gold-blum’s claims would be barred under the pre-AEDPA abuse-of-the-writ doctrine, the AEDPA would not have a genuine retroactive effect and we would affirm the district court’s decision barring his claims under the AEDPA. As we will discuss, we do reach this conclusion.
B. Goldblum’s Arguments
1. Goldblum’s Right to an Evidentiary Hearing
Goldblum’s attorney at the oral argument before us said that all he was asking for was a hearing. Goldblum believes that he is entitled to a district court evidentiary hearing for two reasons: (1) we already found that Goldblum made a “prima facie showing” as to section 2244(b)(2)’s substantive requirements; and (2) the state court did not permit him to develop the record fully. We find these arguments unpersuasive for the reasons that follow.
Goldblum first argues that our determination that he has made a “prime facie showing” under section 2244, thus allowing him to file his second petition, somehow required that the district court hold an evidentiary hearing in making its threshold determination under 28 U.S.C. § 2244(b)(4). Appellant’s br. at 17. More specifically, he states,
This Court, based on specific factual allegations, permitted [Goldblum] to file a successor habeas petition under 28 U.S.C. § 2244(b)(3)(A)(prima facie case that petitioner had either cause and prejudice to excuse failure to raise issues earlier or a demonstration of actual innocence). The district court nevertheless denied Goldblum an evidentiary hearing on these fully supported allegations and, on this incomplete record, ruled that Goldblum had ‘abused’ the writ. *219to file, without reaching the merits of the motion, if the court finds that the movant has not satisfied the requirements for the filing of such a motion. 28 U.S.C. § 2244(b)(4). The movant must get through two gates before the merits of the motion can be considered.
*218Id. at 15.
In reaching this conclusion, Goldblum misunderstands our gatekeeping role in authorizing the filing of second or successive petitions under the AEDPA. Though it well may be that we have made neither the meaning of “prima facie showing” under section 2244(b)(3)(A) nor how that meaning impacts the district court’s section 2244 obligations clear, many other courts of appeals have. Courts often have cited Bennett v. United States, 119 F.3d *219468 (7th Cir.1997), as an instructive opinion in this field. In Bennett, the Court of Appeals for the Seventh Circuit held:
By ‘prima facie showing’ we understand (without guidance in the statutory language or history or case law) simply a sufficient showing of possible merit to warrant a fuller exploration by the district court.9 All that we usually have before us in ruling on such an application, which we must do under a tight deadline (see 28 U.S.C. § 2244(b)(3)(D)), is the application itself and documents required to be attached to it, consisting of the previous motions and opinions in the case. We do not usually have a response from the government, though such a response is authorized. 7th Cir. R. 22.2(c).10 If in light of the documents submitted with the application it appears reasonably likely that the application satisfies the stringent requirement for the filing of a second or successive petition, we shall grant the application. The grant is, however, it is important to note, tentative in the following sense: the district court must dismiss the motion that we have allowed the applicant
Bennett, 119 F.3d at 469-70. At least seven other courts of appeals have adopted this interpretation of “prima facie showing.” See In re Lott, 366 F.3d 431, 432-33 (6th Cir.2004); Williams, 330 F.3d at 281-82; In re Holladay, 331 F.3d 1169, 1173-74 (11th Cir.2003); Bell v. United States, 296 F.3d 127, 128 (2d Cir.2002); Reyes-Requena v. United States, 243 F.3d 893, 898-99 (5th Cir.2001); Thompson v. Calderon, 151 F.3d 918, 925 (9th Cir.1998); Rodriguez v. Superintendent, Bay State Core. Ctr., 139 F.3d 270, 273 (1st Cir.1998), overruled on other grounds by Bousley v. United States, 523 U.S. 614, 118 S.Ct. 1604, 140 L.Ed.2d 828 (1998). Today, we join them in adopting the meaning of “pri-ma facie showing” discussed in Bennett.
In light of these principles, it is clear that Congress did not intend to bind the *220district court in any way by a court of appeals’ preliminary examination of the substantive requirements under section 2244(b)(2), except to the extent that if a court of appeals finds that a petitioner has made a prima facie showing, the district court is obligated to conduct an independent gatekeeping inquiry under section 2244(b)(4). This limited effect of a court of appeals’ initial determination is required because a court of appeals should make its determination within an extremely tight deadline and on the basis of a limited inquiry; thus, it is clear that Congress did not intend that the court of appeals’ preliminary authorization determine how a district court conduct its subsequent analysis. The district court will need to make a more extensive inquiry under section 2244(b)(4) and it “must conduct a thorough review to determine if the motion conclusively demonstrates that it does not meet AEDPA’s second or successive motion requirements,” Reyes-Requena, 243 F.3d at 899 (internal quotation marks and citation omitted). But a court of appeals is obliged only to make a preliminary determination as to whether a petitioner has made a “prima facie showing” with respect to those same requirements, i.e., whether the petition makes “a sufficient showing of possible merit to warrant a fuller exploration by the district court,” Bennett, 119 F.3d at 469. Nevertheless, notwithstanding a district court’s obligation to make an independent gatekeeping inquiry, a district court does not face a requirement that it always conduct an evidentiary hearing in undertaking this more thorough review. Rather, the decision of whether or not to hold an evidentiary hearing is within the distriet court’s discretion. See Schriro, 127 S.Ct. at 1940.
Goldblum also contends that he is entitled to an evidentiary hearing under section 2254(e)(2) because the state court did not permit him to develop the record fully as the court precluded the testimony of two forensic experts who would have provided expert opinions similar to those of Dr. Wecht. Under 28 U.S.C. § 2254(e)(2),
[i]f the applicant has failed to develop the factual basis of a claim in State court proceedings, the court shall not hold an evidentiary hearing on the claim unless the applicant shows that—
(A) the claim relies on — (i) a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court, that was previously unavailable; or (ii) a factual predicate that could not have been previously discovered through the exercise of due diligence; and
(B) the facts underlying the claim would be sufficient to establish by clear and convincing evidence that but for constitutional error, no reasonable factfinder would have found the applicant guilty of the underlying offense.11
Thus, under section 2254(e)(2), if an applicant has developed the factual basis of his claims in the state court, he is not entitled to a federal evidentiary hearing. Furthermore, even if the factual basis is not sufficiently developed, a petitioner must demonstrate that his case falls within the very limited circumstances listed in section 2254(e)(2)(A) and (B), and only then is the district court permitted under the AEDPA, though not required, to grant *221an evidentiary hearing. See Campbell v. Vaughn, 209 F.3d 280, 286-87 (3d Cir.2000). We reiterate that the decision to grant an evidentiary hearing is “left to the sound discretion of district courts.” Schri-ro, 127 S.Ct. at 1939. Additionally, the Supreme Court has made clear that “an evidentiary hearing is not required on issues that can be resolved by reference to the state court record,” as “[i]f district courts were required to allow federal habe-as applicants to develop even the most insubstantial factual allegations in eviden-tiary hearings, district courts would be forced to reopen factual disputes that were conclusively resolved in the state courts.” Id. at 1940 (internal quotation marks and citation omitted). With these principles in mind, we review Goldblum’s evidentiary hearing contention.
We disagree with Goldblum that the factual basis of his claims was not developed sufficiently in the state-court proceedings. The state court conducted a three-day evi-dentiary hearing in which it heard the testimony of Goldblum’s trial attorney, Mr. Rothman, his forensic expert, Dr. Wecht, and the Commonwealth’s rebuttal forensic expert, Mr. Wolson. The matter was on remand to the PCRA court on the sole issue of the potential impact of Dr. Wecht’s testimony on the jury had it been presented at trial. The state court on the remand permitted Dr. Wecht to testify unimpeded with respect to his opinions. The court, however, did not permit testimony from two other experts who Gold-blum planned to introduce to buttress Dr. Wecht’s findings because their testimony exceeded the scope of the remand, although it did admit their affidavits. We conclude that the testimony of Dr. Wecht in conjunction with the affidavits of the proposed experts sufficiently established the factual basis of Goldblum’s claims such that the issues he presented in his habeas corpus petition “can be resolved by reference to the state court record.” Id. at 1940. Thus, Goldblum was not entitled to a district court evidentiary hearing.
Moreover, even if the state court should have regarded the remand as broader in scope so that it permitted the other experts to testify, a conclusion that the state courts did not reach, it is clear that the experts’ opinions would have suffered from the same fatal flaw as Dr. Wecht’s-the lack of photographs or other physical evidence depicting the blood stains would have made their testimony “tentative” and “essentially amounting] to speculation,” as the state court concluded.12 Thus, their testimony likely would have been excluded for, under Pennsylvania law, a court only must entertain expert testimony that would “assist the trier of fact to understand the evidence or to determine a fact in issue,” PA. R. EVID. 702, and relevant evidence “may be excluded if its probative value is outweighed ... by considerations of undue delay, waste of time, or needless presentation of cumulative evidence.” PA. R. EVID. 403. The state standards are similar to those followed in the district courts. Here, after concluding that Dr. Wecht’s findings were unreliable, it would have been completely appropriate for the court to have explained its discretion in precluding cumulative expert testimony on the basis that it would not have been helpful to it in understanding the evidence or determining a fact in issue.13
*222Even if we agreed with Goldblum that he has been denied the opportunity to develop the factual record through no fault of his own, and therefore section 2254(e)(2) did not preclude an evidentiary hearing, such a finding does not necessarily entitle him to one. See Campbell, 209 F.3d at 287. Rather, it merely means that while a hearing is not prohibited under section 2254(e)(2), the district court still retains the discretion to grant a hearing or not. See Schriro, 127 S.Ct. at 1937 (“In cases where an applicant for federal habeas relief is not barred from obtaining an eviden-tiary hearing by 28 U.S.C. § 2254(e)(2), the decision to grant such a hearing rests in the discretion of the district court.”); Campbell, 209 F.3d at 287. “In exercising that discretion, courts focus on whether a new evidentiary hearing would be meaningful, in that a new hearing would have the potential to advance the petitioner’s claim.” Campbell, 209 F.3d at 287. For example, in Campbell we discussed the case of Cardwell v. Greene, 152 F.3d 331, 338 (4th Cir.1998) (internal quotation marks and citation omitted), in which the court held that an evidentiary hearing was permissible under section 2254(e)(2) because it was the state’s fault that the factual record was incomplete, but concluded that it was within the district court’s discretion to deny an evidentiary hearing as the petitioner “ha[d] failed to forecast any evidence beyond that already contained in the record that would help his cause, or otherwise to explain how his claim would be advanced by an evidentiary hearing.”
Likewise, in this case, the magistrate judge did not abuse her discretion in refusing to grant Goldblum a new evidentiary hearing, even if one was permitted under section 2254(e)(2). Goldblum does not have any evidence beyond that already contained in the state-court record that would help his cause. While the state court did not permit two of Goldblum’s experts to testify, their affidavits are part of the state-court record submitted to the district court. Essentially then, Goldblum seeks an evidentiary hearing on the sole ground that he has two more experts, whose affidavits were before the district court, who will present testimony echoing *223that of Dr. Wecht. The “bolstering” testimony of the other experts does not have potential to advance Goldblum’s claims as such testimony cannot overcome the fatal flaw found by the state court, the magistrate judge, and the district court-that absent forensic evidence confirming the distribution of the blood stains, their findings are inconclusive and unreliable. It would not be prudent to hold an evidentiary hearing to reach a conclusion already inevitably reached, and, thus, the magistrate judge certainly did not abuse her discretion in refusing to hear such duplicative testimony on an issue that could readily “be resolved by reference to the state court record.” Schriro, 127 S.Ct. at 1940.
Goldblum makes two additional arguments related to this point. First, Gold-blum contends that “the district court could not fairly adjudicate the claims of actual innocence” because it failed to consider the experts’ affidavits that would confirm and bolster Dr. Wecht’s findings. Appellant’s br. at 23. To the contrary, the Report and Recommendation demonstrates that the magistrate judge did consider the affidavits. See App. at 48 (“Indeed, Dr. Wecht’s affidavit, and the affidavits of other forensic experts, do appear to make a strong case .... ”). However, she agreed with the PCRA court that these affidavits were not probative in light of the finding that the lack of physical evidence depicting the blood stains made their opinions “indeterminate.”
Second, Goldblum believes that the magistrate judge applied “the wrong standard in determining whether Dr. Wecht’s testimony supported a claim of actual innocence” as, according to Goldblum, the testimony need not “unequivocally” or “absolutely” exonerate him, but an expert “need only present opinions to a reasonable degree of certainty.” Appellant’s br. at 25-26. Goldblum has confused the standard governing the admissibility of expert testimony at trial under Federal Rules of Evidence 702 and 703 with the high burden that he must meet to excuse his failure to raise his claim in his first habeas petition under the abuse-of-the-writ doctrine. The district court properly concerned itself only with the latter point.
2. “Cause” under the Abuse-of-the-Writ Doctrine
Goldblum argues as well that the court “did not properly apply the pre-AEDPA abuse of the writ standard.” Id. at 21. The magistrate judge found that Goldblum did not establish “cause” for omission of the blood spatter claim from the first petition because failure to exhaust remedies is not a sufficient excuse. Likewise, she found that the failure to present the claims with respect to the improper instruction on accomplice liability was not excusable as the law governing the instruction was known at the time of the first petition.
Goldblum, however, believes that there is “cause” for failing to present his claims in his first petition so long as the claim was not withheld for “manipulative purposes” or was not “deliberately withheld ... in order to secure an opportunity to pursue unnecessary, and thus vexatious, successive litigation.” Id. He argues that in his case there “was no such ‘manipulative purpose’ or attempt at vexatious litigation” as “[tjhere is no evidence that [he] had actual knowledge of his claim (as neither he nor anyone on his behalf had considered or investigated the claim) or that he deliberately withheld the claim to seek some unfair advantage in a second habeas proceeding.” Id. at 21-22. Moreover, he argues that since his blood spatter claim “would have resulted in a ‘mixed petition’ and thereby would have been subject to *224dismissal pursuant to Rose v. Lundy, 455 U.S. 509, 102 S.Ct. 1198, 71 L.Ed.2d 379 (1982), there was no abuse of the process, and surely no deliberate attempt to gain any litigation advantage.” Appellant’s br. at 22.
We conclude that Goldblum misstates the law. Specifically, the Supreme Court has rejected the “good faith 'deliberate abandonment’ standard” preferred by the dissent in McCleskey, 499 U.S. at 506, 111 S.Ct. at 1477 (Marshall, J., dissenting), and instead has made clear that “[a]buse of the writ is not confined to instances of deliberate abandonment.” Id. at 489, 111 S.Ct. at 1467. The Court concluded that deliberate action is but one example of conduct that would disentitle a petitioner from relief under the abuse-of-the-writ doctrine, stating:
[A] petitioner may abuse the writ by failing to raise a claim through inexcusable neglect. Our recent decisions confirm that a petitioner can abuse the writ by raising a claim in a subsequent petition that he could have raised in his first, regardless of whether the failure to raise it earlier stemmed from a deliberate choice.
Id. at 489, 111 S.Ct. at 1468. Thus, it is clear that the Supreme Court in McCles-key held that a court does not have to find that a petitioner had a “manipulative purpose” or acted with “deliberateness” in withholding a claim from a prior petition in order for the court to reject the petitioner’s contention that he had “cause” under the abuse-of-the-writ doctrine for failing to present the claim in his first petition when he files a subsequent petition. See United States v. Barrett, 178 F.3d 34, 49 (1st Cir.1999) (holding that “[wjhether or not [petitioner’s] failure to [include the claims in the first petition] was intentional is of no moment” under McCleskey); Saahir v. Collins, 956 F.2d 115, 119 (5th Cir.1992) (finding petitioner abused the writ because he “should have known” about the legal theories he failed to advance in his first habeas petition); Campbell v. Blodgett, 997 F.2d 512, 520 (9th Cir.1992) (“To justify this costly litigation strategy [of permitting second habeas petitions], be it deliberate choice, procedurally-constrained decision, or neglect, a petitioner must show cause for the omission and prejudice therefrom.”).
Goldblum’s reliance on the term “manipulative purpose[],” in McCleskey is not meritorious as the term is in no way inconsistent with the Court’s clear holding that deliberateness is not required to establish cause. Put in its proper context, the Court, in discussing the costs of federal collateral litigation, stated, “habeas corpus review may give litigants incentives to withhold claims for manipulative purposes and may establish disincentives to present claims when evidence is fresh.” McCles-key, 499 U.S. at 491-92, 111 S.Ct. at 1469. This statement does not inject a deliberateness requirement into the abuse-of-the-writ jurisprudence, but instead offers one of many examples of conduct that would bar relief.
We similarly reject Goldblum’s argument that we should excuse his failure to include the blood spatter claim in his first habeas corpus petition because the validity of the claim had not been litigated in the state courts and would have resulted in a “mixed petition” under Rose v. Lundy, 455 U.S. 509, 102 S.Ct. 1198, 71 L.Ed.2d 379. We encountered this precise issue in Ben-choff. There, we determined that “Rose v. Lundy requires a petitioner to either fully exhaust all claims prior to filing a petition or to raise both exhausted and unexhaust-ed claims in the first habeas petition.” Benchoff, 404 F.3d at 820. In a case in which an applicant chose the latter course, the petition would be dismissed without *225prejudice and the applicant then properly could refile the petition once all of the claims are exhausted and the refiled petition would not constitute a second or successive petition. Id. Thus, we concluded that failure to exhaust a claim “is not an excuse for [the petitioner’s] failure to raise the claim in his first petition.” Id. at 819; see also Slack, 529 U.S. at 487, 120 S.Ct. at 1605 (“A petition filed after a mixed petition has been dismissed under Rose v. Lundy before the district court adjudicated any claims is to be treated as ‘any other first petition’ and is not a second or successive petition.”). As in Benchoff, Gold-blum’s choice to withhold his blood spatter claim from his first habeas corpus petition while he exhausted it in state court rather than following the procedure prescribed by Rose and its progeny bars this claim under the abuse-of-the-writ doctrine.
3. “Actual Innocence” under the Abuse-of-the-Writ Doctrine
a. The new evidence
Goldblum contends that even if he has not established “cause” and “prejudice” under the abuse-of-the-writ doctrine, he has demonstrated his “actual innocence,” and thus the district court’s dismissal of his second petition was a “miscarriage of justice.” Goldblum faces a very high burden with respect to this assertion. The Supreme Court has instructed us that our authority to excuse the failure to present a claim in the first petition, absent “cause” and “prejudice,” only should be exercised in a “narrow class of cases,” i.e., in “extraordinary instances when a constitutional violation probably has caused the conviction of one innocent of the crime.” McCleskey, 499 U.S. at 494, 111 S.Ct. at 1470. “To establish the requisite probability, the petitioner must show that it is more likely than not that no reasonable juror would have convicted him in light of the new evidence.” Schlup, 513 U.S. at 327, 115 S.Ct. at 867.
In Hubbard v. Pinchak, 378 F.3d 333, 340 (3d Cir.2004), cert. denied, 543 U.S. 1070, 125 S.Ct. 910, 160 L.Ed.2d 805 (2005), we set forth a two-step inquiry a court must take in deciding a claim of actual innocence. First, a court must decide “whether the [petitioner] has presented new reliable evidence ... not presented at trial.” Id. (internal quotation marks and citation omitted). With respect to this inquiry, the Supreme Court has held:
[A] substantial claim that constitutional error has caused the conviction of an innocent person is extremely rare. To be credible, such a claim requires a petitioner to support his allegations of constitutional error with new reliable evidence-whether it be exculpatory scientific evidence, trustworthy eyewitness accounts, or critical physical evidence-that was not presented at trial. Because such evidence is obviously unavailable in the vast majority of cases, claims of actual innocence are rarely successful.
Schlup, 513 U.S. at 324, 115 S.Ct. at 865 (internal citations omitted). Thus, “[without any new evidence of innocence, even the existence of a concededly meritorious constitutional violation is not in itself sufficient to establish a miscarriage of justice that would allow a habeas court to reach the merits of a barred claim.” Id. at 316, 115 S.Ct. at 861.
Second, only if a petitioner first puts forth new evidence not considered by the jury does a court ask “whether it is more likely than not that no reasonable juror would have convicted him in light of the new evidence.” Hubbard, 378 F.3d at 340. In making this second inquiry, a court “must consider all the evidence, old and new, incriminating and exculpatory, without regard to whether it would necessarily *226be admitted under rules of admissibility that would govern at trial,” and “assess how reasonable jurors would react to the overall, newly supplemented record.” House v. Bell, 547 U.S. 518, 126 S.Ct. 2064, 2077, 2078, 165 L.Ed.2d 1 (2006) (internal quotation marks and citation omitted). The Supreme Court repeatedly has emphasized that
[t]he meaning of actual innocence ... does not merely require a showing that a reasonable doubt exists in the light of the new evidence, but rather that no reasonable juror would have found the defendant guilty. It is not the district court’s independent judgment as to whether reasonable doubt exists that the standard addresses; rather the standard requires the district court to make a probabilistic determination about what reasonable, properly instructed jurors would do. Thus, a petitioner does not meet the threshold requirement unless he persuades the district court that, in light of the new evidence, no juror, acting reasonably, would have voted to find him guilty beyond a reasonable doubt.
Schlup, 513 U.S. at 329, 115 S.Ct. at 868. “[T]he petitioner should be allowed to pass through the gateway and argue the merits of his underlying claims” only “if a petitioner ... presents evidence of innocence so strong that a court cannot have confidence in the outcome of the trial unless the court is also satisfied that the trial was free of nonharmless constitutional error.” Id. at 316, 115 S.Ct. at 861.
As applied to the case before us, the precedents we have cited establish that it was Goldblum’s obligation to put forth “new reliable evidence ... not presented at trial,” Hubbard, 378 F.3d at 340 (internal quotation marks and citation omitted), such as “exculpatory scientific evidence, trustworthy eyewitness accounts, or critical physical evidence.” Schlup, 513 U.S. at 324, 115 S.Ct. at 865. Goldblum points to two pieces of evidence that he regards as “new reliable evidence” that demonstrate his innocence: (1) a leading expert in blood spatter evidence, Dr. Wecht, who did not testify at the trial with respect to the blood spatter evidence, has offered his opinion that Miller, and not Goldblum, murdered Wilhelm; and (2) Miller confessed to a member of the State Board of Pardons and the attorney general of Pennsylvania that he participated along with Goldblum in the actual murder. While we have serious doubts whether Dr. Wecht’s opinion constitutes either “new” 14 or “reliable” evidence, or whether Miller’s confession is “reliable,” we will assume for the purposes of this appeal that he has satisfied step one of the two-step “actual innocence” inquiry. Of course, the reliability of the evidence certainly should and will be considered in step two of the analysis in which we must predict the likely impact of the new evidence on the jury.
We find that these circumstances do not present the “extraordinary instance[ ]” where “it is more likely than not that no reasonable juror would have convicted [Goldblum] in light of the new evidence.” Schlup, 513 U.S. at 327, 115 S.Ct. at 867 (emphasis added). This high standard supplies the basis by which we are bound to judge this case and Goldblum’s showing, for the reasons we will discuss, falls short.
*227Goldblum points to seven pieces of evidence presented at the trial which he believes “demonstrated that this was a one-man assault, and the evidence pointed strongly towards Miller.” Appellant’s br. at 9. First, he points to the dying declaration that Wilhelm made to the police wherein he stated, “Clarence Miller did this to me.”15 Second, he distinguishes the small spot found on his shirt cuff that no witness identified positively as blood with the excessive blood found on Miller’s clothes. Third, he believes that the defensive cuts on Wilhelm’s hands and multiple cuts and slashes on his torso, front, back, head and face, show that this was a one-person attack on a moving target. Fourth, he alleges that there was no forensic evidence supporting Miller’s contention that Goldblum began the assault by hitting Wilhelm in the back of his head with a wrench. Fifth, Miller had fresh scratches on his forearms and wrists 14 hours after the homicide while Goldblum did not have any scratches. Sixth, black vinyl gloves recovered from the scene were stained with Wilhelm’s blood and had hairs consistent with Miller’s arm hair. Finally, the blood spatter indicated that the blood was cast off in a left to right movement, suggesting that the killer was in the front passenger seat.16
All of this evidence was presented to the jury. As we have stated, now Goldblum raises two new pieces of evidence which he believes support his claim of innocence: expert testimony supporting his argument to the jury that the blood spatter demonstrated that the killer sat in the front passenger seat, and Miller’s confession to a member of the State Board of Pardons and the attorney general of Pennsylvania that he and Goldblum were involved directly in the murder. Essentially then, the question comes down to this: In the context of a reasonable juror’s review of all of the evidence, is the expert blood spatter evidence and Miller’s confession so persuasive and exculpatory that all 12 members of a jury who voted to convict Goldblum of first-degree murder now would change their minds to the end that it is more likely than not that none of them would vote to convict? 17 The answer to that question is undoubtedly no.
*228We start with our examination of Dr. Wecht’s expert opinion. Dr. Wecht, an expert in forensic pathology, submitted an affidavit wherein he “concluded to a reasonable degree of medical certainty that Mr. Goldblum was not the individual who inflicted the fatal stab wounds to Mr. Wilhelm.” App. at 21. He based this opinion on a number of factors, though predominantly on his finding that the blood spatter on the dashboard demonstrated that the person in the front seat to the right of Wilhelm, i.e., Miller, inflicted the stab wounds. Additionally, he considered the lack of blood found on Goldblum’s clothes in contrast to Miller’s heavily stained clothes, Miller’s blood-stained gloves found at the crime scene, and Wilhelm’s dying declaration.
On direct examination at the PCRA remand hearing, Dr. Wecht repeated his conclusion and elaborated on the bases for his opinion. He opined that it was “highly implausible” that Goldblum inflicted the wounds on Wilhelm given that blood was not found on Goldblum’s clothes. Id. at 129. Additionally, he testified that the gloves found near the scene stained with Wilhelm’s blood contained arm or hand hairs consistent with Miller’s but not Gold-blum’s. Also, Dr. Wecht found relevant that Miller was observed the day after the murder with fresh scratches on his arms and face, but no such scratches were found on Goldblum. Most significantly, Dr. Wecht opined that the blood spatter “cast off,” with the “tails” moving from left to right as found on the dashboard, “are entirely consistent with and buttress and support the conclusion” that Miller inflicted the wounds as he was sitting in the front passenger seat next to Wilhelm, rather than by Goldblum who sat in the rear driver’s-side seat. Id. at 132.
Cross-examination, however, significantly undermined the definitiveness and reliability of Dr. Wecht’s opinion. In fact, Dr. Wecht conceded that he could not “rule out” the possibility that the blood came from Goldblum who got it on his hand when “he was exiting the vehicle.” Id. at 160. Moreover, he testified that he even could not be certain that the spattered blood was Wilhelm’s as it never was tested to ascertain whose blood it was. He also conceded that someone in the back seat could have caused the formation of the blood spatter by reaching over to the front seat. Goldblum, of course, was sitting in the back seat. Dr. Wecht agreed that the circumstance that Goldblum was wearing two layers of clothing could explain the lack of blood found on him.
As we have indicated, the Commonwealth called a forensic consultant, Toby Wolson, to respond to Dr. Wecht’s testimony. Wolson found that because the description of the blood spatter was “limited,” he was prevented from making “a more reliable interpretation of the cause or nature of that particular pattern.” Id. at 212. He opined that there were “other types of situations that may create [the] patterns” that Detective Freeman described besides cast off. Id. at 215. He concluded that there was “[n]ot enough documentation to say which was the most likely cause of [the] pattern,” as the spatter could have come from stab wounds made by someone reaching over the front seat or from defensive hand wounds from the victim. Id. at 218. He opined that, in light of the lack of documentation, no expert could reach any conclusion that was anything more than “hypothetical,” id., and “coming to a conclusion as to who may have done and created the pattern on the dashboard and window is impossible.” Id. at 219. He found that “due to the insufficient documentation,” he could not “tell you if it was the person sitting in the front seat or back seat.” Id. at 226. With respect to the dying declaration, he opined *229that it “is open to interpretation as to what it really means,” and that the hair found on the bloody gloves was not reliable as facial and body hairs, as opposed to head and pubic ham, do not provide for a reliable forensic evaluation. Id. at 228.
In light of this evidence, we are satisfied that Dr. Wecht’s expert testimony regarding the blood spatter would not have had an appreciable, let alone heavy, impact on a reasonable juror in this case for four reasons. First, Goldblum’s trial attorney presented circumstantial evidence regarding the blood spatter suggesting that the person in the front passenger seat, Miller, did the stabbing. Indeed, Goldblum argued at trial that the weapon with which Wilhelm was stabbed was withdrawn in a horizontal motion, splashing a trail of blood horizontally across the dashboard from the left to right towards the passenger seat. We are satisfied that an expert opinion which merely recasts this circumstantial evidence, that was well within the grasp of a reasonable juror without the aid of an expert, would not have a significant influence on a reasonable juror.
Second, the reliability of Dr. Wecht’s opinions is certainly suspect and a reasonable juror would not give enough weight to them to change his or her mind as to Goldblum’s guilt. As Chief Justice Roberts wrote in his dissent in House, “the new evidence is not simply taken at face value; its reliability has to be tested.” House, 126 S.Ct. at 2088 (Roberts. C.J., dissenting). It is clear that the absence of photographs depicting the blood spatter, even if it does not preclude Dr. Wecht from offering an expert opinion, certainly casts doubt on the reliability of that opinion. The Commonwealth’s expert, Mr. Wolson, testified that a number of different scenarios could have caused the blood spatter formation that Detective Freeman described, and, without further documentation, any opinion as to the explanation for the pattern of the blood spatter was mere speculation. Moreover, Dr. Wecht on cross-examination conceded that the evidence of blood spatter did not rule out the Commonwealth’s theory.
Third, Dr. Wecht’s conclusion that Gold-blum did not directly participate in the murder was partially the result of his personal view of non-forensic evidence, such as Wilhelm’s dying declaration, already before the jury. We do not see how Dr. Wecht’s opinion on these particular matters, which in actuality do not seem to require a great amount of scientific expertise (if any at all) and were clearly comprehensible by the jury at the trial, disposi-tively would have influenced them.18
Lastly, the Commonwealth tried the case on the alternate theory that Goldblum was an accomplice to the murder. Because the jury did not make a specific finding as to whether Goldblum was guilty as an accomplice or for committing the actual stabbing, even if a jury believed Dr. Wecht’s speculative testimony on the significance of the blood spatter, we see no reason why it would have reached a different result.19 For these reasons, we con-*230elude that a reasonable juror would not have given significant weight to Dr. Wecht’s speculative expert testimony in the sense that the testimony would have affected the juror’s result.
We now discuss the second piece of new evidence, Miller’s confession. In a deposition on June 7, 1999, Richard Gigliotti, at the time the warden of the Butler County Jail and a member of the Pennsylvania Board of Pardons, testified that on April 29, 1999, about a week before Goldblum’s Board of Pardons’ hearing, he, along with the attorney general of Pennsylvania, met with Miller at the Western Penitentiary and asked him questions about Goldblum’s case. Preliminarily, Miller told them that Goldblum was “the mastermind” behind the land fraud. App. at 326. Miller also mentioned the arson but Gigliotti could not give specific details regarding that aspect of the case. Eventually, Miller told Gi-gliotti that he no longer “denied direct involvement in the actual stabbing of [Wilhelm],” but rather “he now openly admits his involvement along with Goldblum’s involvement in the stabbing and murder of the victim.” Id. at 327-28. Specifically, Gigliotti testified that Miller told him that “Goldblum was sitting in the back seat and inflicted the first wound by coming over the shoulder of ... Wilhelm and stabbing him in, in the lower chest or stomach area.” Id. He also testified that Miller told him “Goldblum definitely started the stabbing and then he took the blade from Goldblum or Goldblum handed him the blade and Miller continued the stabbing,” but Miller did not have a “clear recollection” as to how many times each of them stabbed Wilhelm. Id. at 328-29.
Based on this testimony, Goldblum argues that “[w]hile Miller continues to claim that Goldblum also took part in the stabbing, this dramatic change of testimony is further substantial evidence of the innocence.” Appellant’s br. at 36. We disagree. After all, Miller’s statement to Gi-gliotti, rather than exonerating Goldblum, implicates him in the murder. At the very most, Miller’s statement further impeaches his credibility, but this effect is not significant. When assessing this type of new evidence, we should “consider how the timing of the submission and the likely credibility of the affiants bear on the probable reliability of that evidence.” Schlup, 513 U.S. at 332, 115 S.Ct. at 870. Here, there already was considerable evidence before *231the jury that Miller’s testimony was not credible. Among other impeachment evidence, it was presented to the jury that he lied to the police when he initially denied his involvement and then later admitted that he had lured Wilhelm into the garage. Miller’s different version of the events would not significantly further harm his already strongly impeached testimony. The fact that Miller may have decided to “come clean” more than 20 years after Goldblum’s criminal trial, if he did so, is not persuasive. Moreover, while we do not understand why Miller would have had a motive to lie to Gigliotti, we equally do not understand why he would have a new motive to tell the truth at that late date. Such evidence does not carry great weight.
In sum, mere cumulative impeachment evidence, along with Dr. Weeht’s expert testimony and all of the evidence presented at trial, is not “so strong” that we do not have “confidence in the outcome of the trial” such that “no reasonable juror could have found [Goldblum] guilty.” Schlup, 513 U.S. at 316, 329, 115 S.Ct. at 861, 868 (emphasis added). Applying what Chief Justice Roberts said in his dissent in House, “[t]he question is not whether [Goldblum] was prejudiced at his trial because the jurors were not aware of the new evidence, but whether all the evidence, considered together, proves that [Goldblum] was actually innocent, so that no reasonable juror would vote to convict him.” House, 126 S.Ct. at 2087 (Roberts, C.J., dissenting). While Goldblum may have been prejudiced by his failure to present this new evidence, even with it he simply cannot meet the stringent burden that applies to our review.
A comparison of two recent cases, the Supreme Court opinion in House, 126 S.Ct. 2064, and our opinion in Albrecht v. Horn, 485 F.3d 103 (3d Cir.2007), reinforces our conclusion. In House, 126 S.Ct. at 2068, the defendant was convicted of the murder of Carolyn Muncey in spite of his contention at trial that Mrs. Muncey’s husband committed the murder. The results of a testing that the FBI ran on Mrs. Muncey’s nightgown and panties and defendant’s jeans were central to the prosecutor’s case. Id. at 2072. Specifically, at trial the prosecution presented evidence that semen consistent with the defendant’s was found on Mrs. Muncey’s nightgown and panties, and small bloodstains consistent with her blood were found on the defendant’s jeans. The prosecution argued, and the jury apparently believed, that the defendant committed or attempted to commit a sexual assault on Mrs. Muncey and killed her when she resisted.
The defendant sought habeas corpus relief asserting claims of ineffective assistance of counsel and prosecutorial misconduct. Id. at 2075. The government argued that the claims were procedurally defaulted because the defendant did not raise them in his first post-conviction proceeding in the state court. The defendant responded that he had new evidence that undermined the semen and blood evidence the prosecution introduced at trial, and thus, his procedural default should be excused under the “actual innocence” theory. The Supreme Court agreed with the defendant.
The defendant introduced three key pieces of new evidence. First, he presented new DNA testing showing that the semen found on Mrs. Muncey’s nightgown and panties came from her husband and not from him. Id. at 2078-79. The Court concluded that this new evidence cast doubt on the sexual motive presented by the prosecution at trial and removed key physical evidence linking the defendant to the crime. Id. at 2079.
Second, the defendant presented the expert testimony of a forensic pathologist *232who testified that the blood found on the defendant’s pants “was chemically too degraded, and too similar to blood collected during the autopsy, to have come from Mrs. Muncey’s body on the night of the crime.” Id. at 2080. The defendant also presented evidence revealing that the blood vials from the autopsy were packed in a cardboard box with the defendant’s pants during the transport of the evidence to the FBI, in violation of proper procedure, and that roughly a vial and a half of the blood samples spilled during the transport. Id. The Court found that “the evi-dentiary disarray surrounding the blood” along with the expert’s testimony “would prevent reasonable jurors from placing significant reliance on the blood evidence.” Id. at 2083.
Third, the defendant supplied evidence from people who knew the Munceys, suggesting that the other main suspect in the case, Mr. Muncey, regularly abused his wife, as well as testimony that Mr. Muncey confessed to the murder around the time of the defendant’s trial and that Mr. Mun-cey had the opportunity to commit the crime. Id. at 2083-84. The Court found that while “[t]he evidence pointing to Mr. Muncey is by no means conclusive” and “[i]f considered in isolation, a reasonable jury might well disregard it,” “[i]n combination ... with the challenges to the blood evidence and the lack of motive with respect to [the defendant], the evidence pointing to Mr. Muncey likely would reinforce other doubts as to [the defendant’s] guilt.” Id. at 2085.
Based predominantly on these three pieces of evidence, the Court concluded that while “[t]his is not a case of conclusive exoneration” and some of the evidence “still supports] an inference of guilt,” “the central forensic proof connecting [the defendant] to the crime-the blood and the semen-has been called into question, and [the defendant] has put forward substantial evidence pointing to a different suspect.” Id. at 2086. As we pointed out in Albrecht, with respect to House, 485 F.3d at 125, “[t]he new DNA evidence effectively destroyed the theory of rape as the motive for the murder,” and “[without the blood evidence, [the defendant] did not have a motive, but the victim’s husband did.” Thus, the Supreme Court found that “although the issue is close,” this was “the rare case where, had the jury heard all the conflicting testimony, it is more likely than not that no reasonable juror would lack reasonable doubt.” House, 126 S.Ct. at 2086.
In contrast to House, in Albrecht, 485 F.3d at 120-21, we found that a defendant, who was convicted of first-degree murder, two counts of second-degree murder, and arson for causing the death of his wife, mother, and daughter by setting the family home on fire, did not present new evidence establishing his “actual innocence” that would have permitted him to proceed with his habeas corpus petition despite the forfeiture of his state habeas corpus claims on procedural grounds. At trial, the crux of the prosecution’s case was that an act of arson caused the fire and the identity of the arsonist could be inferred from the violence and hostility the defendant had directed toward his wife in the months before the fire, including recent threats of burning down the house. As evidence that the fire was the product of arson, the prosecution introduced the testimony of a fire expert who opined that based on the char patterns, the arsonist started the fire by igniting gasoline that had been poured on the floor. The defendant, on the other hand, while admitting that he abused his wife, argued that a cigarette left to smolder in an upholstered chair accidentally started the fire.
*233The defendant sought habeas corpus relief, arguing “that new developments in fire science prove his claim of actual innocence,” thereby excusing the procedural default of his state habeas corpus claims. Id. at 120. In particular, the defendant presented the testimony of a fire protection engineer who opined that “all of the observations relied on by [the prosecution’s fire expert] to support his conclusion that the fire was set in the kitchen using gasoline ... are now understood to be equally consistent with an accidental fire that resulted in full room involvement.” Id. at 120-21. Thus, he believed “that the fire could have been accidental in origin,” as “fire-scene evidence of an accidental fire that has reached full room involvement is indistinguishable from the evidence seen after an incendiary fire that likewise affected the full room.” Id. at 124.
A critical issue before us was whether “in light of [the] new evidence ... ‘it is more likely than not that no reasonable juror would have found petitioner guilty beyond a reasonable doubt.’ ” Id. at 121 (quoting Schlup, 513 U.S. at 327, 115 S.Ct. at 867). We held that the defendant had not satisfied his burden. First, we pointed out that the defense expert’s opinion did not rule out the possibility that the fire was set intentionally, as concluded by the prosecution’s expert at trial.
Second, we compared the case to the circumstances before the Supreme Court in Hotise. In House, the new evidence undermined the original evidence pointing to the identity of the murderer and the motive for the murder. On the other hand, the new evidence in Albrecht merely “raised questions about the alleged incendiary nature of the fire, while identity and motive ... were established by other evidence,” and “[t]he substantial remainder of the Commonwealth’s case has not been discredited and provides ample evidence of guilt.” Id. Specifically, the defense expert’s testimony “did nothing to undermine the Commonwealth’s damaging evidence of [the defendant’s] pattern of hostility and violence toward [his wife], his attempt to purchase gasoline to put in a can the day before the fire, the immediate discovery of the empty hydraulic oil can in the trunk of his car that tested positive for gasoline, and his numerous threats to burn down the house and do further harm to his wife.” Id.
Finally, we found that the testimony of the defendant’s new expert did not carry much weight because his testimony was similar to the trial testimony of another defense expert that the jury apparently rejected, and even of the testimony of the prosecution’s expert, who though he testified at trial that the fire was ignited with gasoline, on extensive cross-examination testified as to some facts that supported the defense’s theory. Based on this analysis of the testimony, we found that the defendant “cannot exploit the new scientific knowledge here, assuming for the sake of argument that it is new, because of ample other evidence of guilt” such that we could not “conclude that, had the jury heard all the conflicting testimony, it is more likely than not that no reasonable juror viewing the record as a whole would lack reasonable doubt.” Id. at 126.
The teaching of House and Albrecht surely is that an actual innocence inquiry is fact intensive and we painstakingly have made such an inquiry here. We find that the new evidence put forth in this case is much more in line with the evidence in Albrecht than the evidence in House. In House, the new scientific blood and semen evidence removed the alleged sexual motive presented to the jury at trial and suggested that the defendant was not even at the scene of the murder, just as the defense argued at trial. As the Supreme *234Court concluded, “the central forensic proof connecting [the defendant] to the crime ... has been called into question.” House, 126 S.Ct. at 2086. The case before us is significantly different. In our case, the blood spatter evidence and Miller’s confession do not detract from Goldblum’s motive for the murder, as it is clear that he still had a motive to relieve himself of the debt due Wilhelm and eliminate Wilhelm’s threats of reporting his illegal activities to the authorities. Additionally, the new evidence does not remove Goldblum from the scene of the murder. In fact, Miller’s confession puts him there as a participant which, of course, is what the prosecution contended at the trial.
In contrast to the new evidence in House, the new evidence in our case is materially indistinguishable from the evidence in Albrecht. First, the evidence here and in Albrecht did not eliminate the possibility that the crime occurred as set forth by the prosecution at trial. Like the fire expert’s opinion in Albrecht that did not eliminate the possibility that the defendant intentionally set the fire as the prosecution argued at trial, Dr. Wecht similarly conceded that his opinion left open the possibility that Goldblum, who sat in the back seat, could have done the stabbing.
Second, like the expert testimony in Al-brecht, Dr. Wecht’s testimony would not carry significant weight with the jury. The defense in this case put forth strong circumstantial evidence at trial that the blood spatter demonstrated that the front-seat passenger did the stabbing and it argued for adoption of this theory to the jury. There is no indication that expert testimony on this point would lead to a different conclusion. Additionally, as demonstrated by the cross-examination of Dr. Wecht at the PCRA hearing and the convincing testimony of the Commonwealth’s expert, Mr. Wolson, Dr. Wecht’s opinions are subject to significant impeachment grossly undermining his conclusions. We are convinced that a reasonable juror would reach the same conclusion that we reach.
Finally, like the evidence in Albrecht with respect to the facts of that case, the new evidence here does not undermine evidence of Goldblum’s motive to kill Wilhelm, nor does it remove him from the scene of the murder. At the very most, if the jury had believed the new evidence, while there would have been a question how Wilhelm’s killing unfolded, the new evidence would not have absolved Gold-blum from the commission of the crime. After all, he initiated the scheme to lure Wilhelm to the parking garage, was present at the murder scene, had a strong motive to kill Wilhelm to cover up other crimes that he had committed and eliminate the debt he owed, conspired with Miller to create an alibi, and attempted to hire an assassin to kill Miller, the prosecution’s main eyewitness, after being arrested for the murder. The new evidence did nothing to undermine this very damaging evidence. In light of the foregoing, we simply cannot conclude that no reasonable juror would vote to convict Goldblum even considering the new evidence.
b. The jury instruction on accomplice liability
There is yet one more issue on this appeal. The Supreme Court in Schlup stated that the actual innocence standard requires the court “to make a probabilistic determination about what reasonable, properly instructed jurors would do.” Schlup, 513 U.S. at 329, 115 S.Ct. at 868 (emphasis added). In this regard, Gold-blum asserts in his habeas corpus petition that his trial attorney was ineffective for failing to object to the jury charge on accomplice liability. While, as we already *235have explained, we do not make an ultimate determination of the merits of the case, we must evaluate the instruction to decide if it was erroneous, and assuming it to have been incorrect, if its correction, in conjunction with the allegedly new evidence, would change the jurors’ minds as to Goldblum’s guilt.
The trial court judge gave the following instruction concerning accomplice liability:
Now running through this case is the legal question of an accomplice. A person may be guilty of a crime although he did not commit the specific offense. This occurs when he acts as an accomplice of the one who commits the crime. A person is an accomplice or legally accountable for the conduct of another when he: 1. With intent of promoting or facilitating the commission of a crime he aids or agrees or attempts to aid such other person in planning or committing the crime or he solicits such other person to commit the crime.
Now, in order to find a person guilty as an accomplice, the evidence must be established beyond a reasonable doubt of a shared criminal intent with the other person. That is, both persons must intend that the criminal act occur.
Now, what does that mean? By ‘shared criminal intent’ we mean that both men, to be found guilty, even though only one committed the act, both intended that the specific crime, which was completed, was intended. An example follows: If two men specifically agree that they are going to kill another man, and one man holds him and the other man shoots him, they are both guilty of first degree murder because one was an accomplice in the holding and the man who fired the shot, of course, is guilty of first degree murder because he inflicted the wound.
Another example of accomplice, in the law, is this: If two men agree to rob a bank and one man stands outside as the lookout and another man goes in, and the man who is inside the bank shoots and kills a teller, they are both guilty of first degree murder, including the man standing out front, because he was an accomplice who was actively aiding the man who was inside the bank performing the robbery and both men should be guilty of first degree murder because of a shared criminal intent to commit that specific crime.
Now, an accomplice is one who knowingly and voluntarily cooperates or aids another in committing a crime. He is not merely a passive bystander who happens to observe an illegal act and does not participate in it. Nor is he someone who sees a crime being committed and fails to report it to the police. Instead, an accomplice is someone who knowingly, voluntarily and purposely joins with someone else in the performance of that specific crime.
Now, the Commonwealth of Pennsylvania contends that the defendant, Charles Goldblum, either killed George Wilhelm himself or was an accomplice with Clarence Miller in that killing.
Most significantly the judge later instructed the jury on application of accomplice liability to the specific facts of the case:
In addition, if you find that evidence established beyond a reasonable doubt that Charles Goldblum was an accomplice with Clarence Miller in the intentional killing of George Wilhelm, you may find the defendant guilty of first degree murder. In that event, you have to conclude that both Goldblum and Miller intended to kill George Wilhelm when they took him up there [to the parking garage] and that Goldblum was *236an accomplice with Miller in the killing of Mr. Wilhelm. So, in that event, if you find that the defendant intentionally killed George Wilhelm, you may find him guilty of first degree murder. If you find that he was an accomplice in such a killing, as I have defined that, you may also. On the other hand, if you find that the evidence has failed to establish beyond a reasonable doubt that George Wilhelm on the night in question, or failed to establish that Charles Goldblum was an accomplice with Clarence Miller in the intentional killing of George Wilhelm, you must find the defendant not guilty of first degree murder. (Emphasis added.)
Goldblum believes the instruction was “plainly erroneous” as it did not require the Commonwealth to prove that Goldblum had the specific intent to kill, which is an element of first-degree murder under Pennsylvania law. Appellant’s br. at 31-32. Specifically, he takes issue with (1) the “shared intent” language in the first instruction which he contends misled the jury into believing that he need not have his own specific intent to kill, i.e., the flaw in the robbery example, and (2) the suggestion in the second instruction that a finding that he is an accomplice compels the jury to conclude that he had the specific intent to kill, which he argues “places the proverbial cart before the horse.” Id. at 33-35. The Commonwealth believes that, overall, the instruction was accurate and that any error in it was overcome by the charge as a whole. In this regard, the Commonwealth argues that the instruction, when read in its entirety, made it clear that the jury needed to .find that Goldblum possessed the intent to murder Wilhelm in order to find him liable as an accomplice.
Laying aside the fact that Goldblum does not contend that his trial attorney objected to this aspect of the charge at the trial, and the inference that we can draw from the absence of an objection that he saw no prejudice from the charge, we independently agree with the Commonwealth. It is well-settled that under Pennsylvania law a defendant cannot be convicted of first-degree murder as an accomplice “unless the Commonwealth proves that he harbored the specific intent to kill.” Smith v. Horn, 120 F.3d 400, 410 (3d Cir.1997). “The Commonwealth need not prove that the defendant actually performed the killing, but it must prove he intended for the killing to occur.” Id. (internal citations omitted). “In considering whether the jury instruction in this case adequately convey[s] this critical feature of Pennsylvania homicide law, we focus initially on the language that is claimed to be. erroneous, but we must review this portion of the instructions in the context of the charge as a whole.” Bronshtein v. Horn, 404 F.3d 700, 710 (3d Cir.2005) (internal quotation marks, citation, and footnote omitted), cert. denied, 546 U.S. 1208, 126 S.Ct. 1320, 164 L.Ed.2d 115 (2006). “The proper inquiry is whether there is a reasonable likelihood that the jury has applied the challenged instructions in a way that violates the Constitution.” Id. Thus, here we must determine when reviewing the challenged portion of the instructions in the context of the charge as a whole, whether there is a reasonable likelihood that the jury convicted Goldblum of first-degree murder without finding beyond a reasonable doubt that he intended the killing of Wilhelm.
Goldblum directs us to two cases: Smith, 120 F.3d 400, and Laird v. Horn, 414 F.3d 419 (3d Cir.2005), cert. denied, 546 U.S. 1146, 126 S.Ct. 1143, 163 L.Ed.2d 1014 (2006), both arising from state court convictions. We decided in both cases that the jury charge on accomplice liability lacked the critical instruction that the jury *237must find that the defendant had the specific intent to kill. Those cases, however, are clearly distinguishable from the one before us.
In Smith, Smith and another individual, Alston, entered a pharmacy with the intention of robbing it, and in the course of the robbery, a fatal gunshot wound was inflicted to the head of a person named Sharp. Smith, 120 F.3d at 404. Smith was tried on the homicide charge. Id. The jury instructions on homicide included the following language:
[I]f ... you find that Smith and Alston were accomplices of each other, then it is not important for you to determine which one actually pulled the trigger that brought about the killing of Richard Sharp, if you find beyond a reasonable doubt that one of the two did so and were [sic] acting as the accomplice of each other at the time. In order, however, to find Clifford Smith to be guilty, you need not conclude, as I said, that he was the actor; that is, if I can use the word ‘shooter,’ but he was, nevertheless, acting as an accomplice of Alston and it was his intent of promoting or facilitating that act and the killing was done in furtherance of the robberies, if you so find, then he would be guilty as though he were the actual perpetrator ....
Id. at 405 (emphasis added). The court continued:
If, and I emphasize this, you find that one was the accomplice of the other and that one of the two actually performed the killing, you, the jurors, need not agree on the role or roles played by the respective parties; that is, by this defendant and his accomplice, if you find that that was the position of both, provided that each of you is satisfied that the crime was actually perpetrated by the defendant or by the accomplice of the defendant.
Id. (emphasis added). The court then instructed the jury on first-degree murder:
[T]he elements of first degree murder are the unlawful killing of another person done intentionally ... plus malice .... If these elements have been established beyond a reasonable doubt, you may, on the theory that one was the perpetrator and the other the accomplice, find Clifford Smith guilty of murder in the first degree ....
Id. Relying on these instructions, the jury convicted Smith of the first-degree murder charge. Id. at 406.
The case came before us on an appeal after the district court denied a petition for habeas corpus. Smith argued that “the jury was incorrectly instructed that if it found beyond a reasonable doubt that one of the men had the specific intent to kill, and that Smith intended to commit the robbery that resulted in the killing, this would be sufficient to convict Smith of first-degree murder.” Id. at 409. We agreed with Smith, finding that “[a] fair reading of the jury instructions ... permitted the jury to convict Smith of murder in the first degree without first finding beyond a reasonable doubt that Smith intended that Sharp be killed.” Id. at 411. Specifically, we were concerned with the fact that the charge repeatedly used the word “accomplice” in isolation, without indicating whether the trial court was using the word in reference to the robbery, the killing, or both. Id. at 411-12. “The charge thus blurred the distinction between ‘accomplice in the robbery’ and ‘accomplice in the killing,’ leading the jury to believe that an accomplice for one purpose is an accomplice for all purposes ... [and] allowed Smith to be convicted of first-degree murder if the jury found that either he or his accomplice in the robbery intended to kill Sharp.” Id. at 412.
*238Likewise, in Laird, Laird and a eo-de-fendant met a third man at a bar, following which the three men left together shortly after the bar closed. Laird, 414 F.3d at 421. The third man was found dead the following evening. Id. Laird and the co-defendant jointly were tried for the murder, as well as related charges including kidnaping, aggravated assault, unlawful restraint, false imprisonment, conspiracy, and possession of an instrument of crime. Id. at 422. While both admitted to their participation in the kidnaping and being present when the victim was killed, they pointed their fingers at each other claiming that the other defendant inflicted the fatal wounds. Id. at 422, 426.
At trial, the court gave an instruction on accomplice liability which included the charge that a defendant
is an accomplice ... if with the intent of promotion or facilitating commission of a crime he solicits, or commands or encourages or requests the other person to commit it or if he aids, agrees to aid, or attempts to aid the other person in planning the crime or committing the crime .... You may find the defendant guilty of a particular crime on the theory that he was an accomplice so long as you are satisfied beyond a reasonable doubt that the crime was committed and the defendant was an accomplice of the person who committed it.
Id. at 426. The court then gave the following instruction on first-degree murder:
You may find a defendant guilty of first degree murder if you are satisfied that the following four elements have been proved beyond a reasonable doubt: First, that Anthony Milano is dead. Second, that a defendant or an accomplice of the defendant killed him. Third, that the killing was with specific intent to kill. And, fourth, that the killing was with malice as I have defined that term for you. A killing is with specific intent to kill if it is willful, deliberate, and premeditated; that is, if it is committed by a person who has a fully informed intent to kill and is conscious of his own intent.
Id.
Because both Laird and the co-defendant admitted to participating in the kid-naping but denied having the intent to kill the victim or helping the other kill him, Laird argued in his habeas corpus petition that the instruction on accomplice liability, which consistently referred only to “an accomplice” of “a crime,” allowed the jury to convict him of first-degree murder even though he did not possess an intent to kill. Id. We found the case indistinguishable from Smith, concluding that “[gjiven the court’s instruction on accomplice liability, the jury could easily have convicted Laird of first degree murder based on his conspiring with [the co-defendant] to kidnap or assault [the victim] even if jurors were not convinced beyond a reasonable doubt that Laird intended to kill him.” Id. at 427.
The facts before us and the instruction on accomplice liability are clearly distinguishable from both those in Smith and Laird. In those cases, we found the instructions on accomplice liability constitutionally inadequate because they allowed the jury to conclude that, as we put it in Smith, 120 F.3d at 412, “an accomplice for one purpose is an accomplice for all purposes.” Such an inference in those cases was quite misleading because the defendants either admitted or the evidence was strong that they participated in a related crime that immediately preceded the killings, i.e., Smith in the robbery and Laird in the kidnaping and assault. Thus, we were concerned that the failure of the court to refer specifically to the killing when discussing accomplice liability al*239lowed the jury to conclude that a finding of accomplice liability on the related crime compelled a finding of accomplice liability on the hilling. Such a conclusion would not have been correct for a conviction for first-degree murder as an accomplice requires the Commonwealth to prove that a defendant “harbored the specific intent to kill.” Smith, 120 F.3d at 410.
This appeal does not present a similar problem. It is clear here that the trial court’s instruction on accomplice liability refers only to the killing of Wilhelm and not to a related crime. This conclusion is inescapable for two reasons. First, the Commonwealth did not allege that Gold-blum committed a crime immediately before the killing, such as robbery or kid-naping, although it did charge that he committed crimes much earlier. Thus, the jury could not have drawn a reasonable inference allowing it to tag Goldblum as an accomplice in a related crime and then improperly extended that finding to Wilhelm’s killing.
Second, contrary to the instructions in Smith and Laird, the instruction here on accomplice liability specifically stated that Goldblum was charged as an accomplice to the murder and did not use the word “accomplice” in isolation. For instance, after the state court gave the general instruction on accomplice liability it stated, “Now, the Commonwealth of Pennsylvania contends that the defendant, Charles Gold-blum, either killed George Wilhelm himself or was an accomplice with Clarence Miller in that killing.” Likewise, as we already have indicated, the judge later instructed the jury on application of accomplice liability to the specific facts of the case, as follows:
In addition, if you find that evidence established beyond a reasonable doubt that Charles Goldblum was an accomplice with Clarence Miller in the intentional killing of George Wilhelm, you may find the defendant guilty of first degree murder. In that event, you have to conclude that both Goldblum and Miller intended to kill George Wilhelm when they took him up there [to (the parking garage]) and that Goldblum was an accomplice with Miller in the killing of Mr. Wilhelm. So, in that event, if you find that the defendant intentionally killed George Wilhelm, you may find him guilty of first degree murder. If you find that he was an accomplice in such a killing, as I have defined that, you may also. On the other hand, if you find that the evidence has failed to establish beyond a reasonable doubt that George Wilhelm on the night in question, or failed to establish that Charles Goldblum was an accomplice with Clarence Miller in the intentional killing of George Wilhelm, you must find the defendant not guilty of first degree murder. (Emphasis added.)
It is difficult to understand how the court could have made the charge clearer on the required intent element of the case and, in the circumstances, we have no doubt that a reasonable juror would understand that he must find that Goldblum intended for the killing to occur, as opposed merely to finding that Goldblum had an intent to commit some related crime, in order to find Gold-blum liable as an accomplice to the murder.
In the end, we find in considering the charge as whole as, of course, we should do, there is no reasonable likelihood that the jury convicted Goldblum of first-degree murder without finding beyond a reasonable doubt that he intended that Wilhelm be killed. While the robbery example in the charge was incorrect, we conclude that the remainder of the charge was entirely accurate and sufficiently explained to the jury that it had to find that *240Goldblum intended the killing to find him liable under the theory of accomplice liability. The robbery example could not mislead the jury because, as explained above, the case had nothing to do with a robbery or any other related crime that immediately proceeded the killing.
We recognize that if the facts in this case were akin to those in Smith and Laird, we might have reached a different result based on this instruction. But we are obligated to decide the case based on the facts before us. We emphasize that the trial court repeatedly told the jury that it needed to find that Goldblum intended to kill Wilhelm if it was to find him liable as an accomplice. For'these reasons, we do not conclude that Goldblum’s argument-that the instructions put the “cart before the horse”-is convincing. In reviewing the charge as a whole, we conclude that there is not a reasonable likelihood that the jury convicted Goldblum as an accomplice without first finding that he had a specific intent to kill.
Lastly, Goldblum takes issue with the “shared intent” language which he contends misled the jury into believing that he need not have his own specific intent to kill. Goldblum asserts that “shared intent” suggests that intent of one actor may be attributed to another. Such an interpretation contradicts the reasonable understanding of that phrase, as well as the court’s specific instructions informing the jury otherwise. The court could not have made it any more obvious that “a shared criminal intent with the other person” means that “both persons must intend that the criminal act occur” and that “both men, to be found guilty, even though only one committed the act, both intended that the specific crime, which was completed, was intended.” Additionally, we on several occasions have held that a properly formulated “shared intent” charge accurately stresses the need to find a specific intent for the purposes of accomplice liability, see Bronshtein, 404 F.3d at 711-12; Everett v. Beard, 290 F.3d 500, 513 (3d Cir.2002), as has the Supreme Court of Pennsylvania, see, e.g., Commonwealth v. Cox, 581 Pa. 107, 863 A.2d 536, 550-51 (2004); Commonwealth v. Rife, 454 Pa. 506, 312 A.2d 406, 408-09 (1973); Commonwealth v. Wilson, 449 Pa. 235, 296 A.2d 719, 721 (1972).
In sum, based on this record, we cannot conclude that no reasonable juror would not have voted to convict Goldblum of first-degree murder in light of the new evidence, even if the instructions on accomplice liability had been completely accurate. Plainly put, there is just too much evidence here establishing Goldblum’s guilt and, overall, the charge was not misleading. In particular, there is convincing evidence that Goldblum and Miller orchestrated a plan to lure Wilhelm to the parking garage. Goldblum admitted to being at the crime scene. He had a motive to kill Wilhelm inasmuch as Wilhelm knew of Goldblum’s involvement in the land fraud scheme and the arson, and demanded payment of his debt. Moreover, after the assault Goldblum and Miller agreed to lie to the police in order to provide each with an alibi. They then left the scene together. After being taken into custody, Gold-blum sought to hire an assassin to kill Miller, the Commonwealth’s chief witness. In light of this evidence, we simply cannot conclude that the new evidence would have led any juror not to vote to convict Gold-blum. For this reason, Goldblum has not established his “actual innocence” and the district court correctly denied his second application for a writ of habeas corpus.
V. CONCLUSION
For the foregoing reasons, we will affirm the district court’s order of December *24113, 2005, denying the petition for a writ of habeas corpus.
. There is not the slightest suggestion in the record that the senator or anyone on his staff had any involvement in the scheme.
. The allegations relating to the plan to murder Miller were not part of Goldblum’s subsequent indictment. It is unclear from the record if the Commonwealth ever charged Goldblum in the scheme to murder Miller. Moreover, the assistant district attorney at oral argument was unaware if the district attorney had brought charges against Gold-blum in the Miller scheme or, if there had been such charges, whether Goldblum had been convicted of them.
. Goldblum's trial attorney wanted to argue that Goldblum was, at most, guilty of involuntary manslaughter based on his presence at the scene of the murder and his failure to prevent Miller from stabbing Wilhelm. Though the trial judge agreed that that argument advanced a viable defense, Goldblum did not want to pursue it and, on the record, waived the presentation of the defense.
. At a separate trial, a jury convicted Miller of first-degree murder as an accomplice.
. Dr. Wecht was the coroner of Allegheny County when Wilhelm was murdered in 1976 but he was not involved in this case at the time of the murder and did not supervise it.
. We will discuss the testimony of Dr. Wecht in greater detail later. See infra at 228-30.
. Under section 2244(b)(4), the court must dismiss the claims in the second petition unless they meet the substantive standard under section 2244(b), which states:
(b)(1) A claim presented in a second or successive habeas corpus application under section 2254 that was presented in a prior application shall be dismissed.
(2) A claim presented in a second or successive habeas corpus application under section 2254 that was not presented in a prior application shall be dismissed unless-(A) the applicant shows that the claim relies on a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court, that was previously unavailable; or
(B)(i) the factual predicate for the claim could not have been discovered previously through the exercise of due diligence; and (ii) the facts underlying the claim, if proven and viewed in light of the evidence as a whole, would be sufficient to establish by clear and convincing evidence that, but for constitutional error, no reasonable factfin-der would have found the applicant guilty of the underlying offense.
. We have held that the AEDPA gatekeeping scheme did more than "simply put in statutory form what McCleskey had already defined as abuse of the writ.” Zayas, 311 F.3d at 257. Instead, rather than supplanting the abuse-of-the-writ doctrine, to which the Supreme Court has referred as "a complex and evolving body of equitable principles informed and controlled by historical usage, statutory developments, and judicial decisions,” Felker v. Turpin, 518 U.S. 651, 664, 116 S.Ct. 2333, 2340, 135 L.Ed.2d 827 (1996), the AEDPA built on the doctrine. See Zayas, 311 F.3d at 257. In this regard, "the abuse of writ doctrine retains viability as a means of determining when a petition should be deemed ‘second or successive’ under the statute,” Benchoff, 404 F.3d at 817, and vice versa inasmuch as the terms of the AEDPA inform judicial consideration with respect to abuse-of-the-writ inquiries. See Zayas, 311 F.3d at 257.
. "[S]ufficient showing of possible merit” in this context does not refer to the merits of the claims asserted in the petition. Rather, it refers to the merits of a petitioner's showing with respect to the substantive requirements of 28 U.S.C. § 2244(b)(2). The merits of the claims in a second petition may not be considered by the district court until the application clears the "two gates” erected under section 2244, that of the court of appeals and that of the district court. In In re Turner, 267 F.3d 225, 228 n. 2 (3d Cir.2001), we suggested that under Bennett an applicant may have to make a "prima facie showing” that his underlying habeas corpus claim has merit during the pre-filing authorization stage. We did not perceive fully at that time what the court meant in Bennett, as other courts subsequently have explained that case, and today we conclude that the "prima facie showing” under section 2244 refers merely to the pre-filing substantive requirements of section 2244(b)(2). See Williams, 330 F.3d at 282 (“[T]he ‘showing of possible merit’ alluded to in Bennett relates to the possibility that the claims in a successive application will satisfy the stringent requirements for the filling of a second or successive petition, not the possibility that the claims will ultimately warrant a decision in favor of the applicant.”) (internal quotation marks and citations omitted).
. We have similar rules. See 3d Cir. L.A.R. 22.5(a) (stating that an application for authorization to file a second or successive petition must be accompanied by the proposed new petition, all prior petitions; copies of the docket entries; copies of all magistrate judge reports, district court opinions, and orders disposing of prior petitions; and any other relevant documents); 3d Cir. L.A.R. 22.5(d) ("Any response to the application must be filed within 7 days of the filing of the application with the clerk.”).
. While an argument can be made that section 2254(e) should not apply at all in these circumstances because section 2244 requires a mere threshold determination that does not involve the merits of the claims, section 2254(e) makes clear that it applies in all proceedings "instituted by an application for a writ of habeas corpus by a person in custody pursuant to the judgment of a State court.” A determination under section 2244, while not on the merits of the claims, falls within this language.
. Of course, on appeal the Superior Court upheld the state trial court’s limitation of the scope of the remand. We address the point from an inquiry as to whether the record was developed sufficiently for our analysis. We certainly do not review the conclusion of the state courts on state law.
. In a related point, Goldblum argues that because he was not permitted to develop the *222state-court record fully, the magistrate judge erred by applying the presumption of correctness to the state court's factual findings under 28 U.S.C. § 2254(e)(1), particularly in reaching the conclusion that Dr. Wecht's testimony was speculative. Under 28 U.S.C. § 2254(e)(1),
In a proceeding instituted by an application for a writ of habeas corpus by a person in custody pursuant to the judgment of a State court, a determination of a factual issue made by a State court shall be presumed to be correct. The applicant shall have the burden of rebutting the presumption of correctness by clear and convincing evidence.
As discussed above, we find that Goldblum sufficiently developed the state-court record, and, thus, we find no error in the magistrate judge's decision to apply the presumption of correctness to the state court's factual findings.
Nevertheless, in an attempt to rebut this presumption, Goldblum’s attorney emphasized at oral argument that the Commonwealth "conceded” in its Answer to Gold-blum’s “Motion for Leave of Court to Serve Respondents with Request for Production of Documents,” dated October 29, 2004, that Detective Freeman provided a "detailed description” of the blood spatter. We, however, do not see how this is relevant as the Commonwealth’s categorization of the evidence is not binding on the views of the experts or the court. Detailed or not, the Commonwealth's expert and the court found that the officer's recollection was not sufficient evidence on which an expert could base his conclusion, and in the absence of photographs, the opinion is unreliable. The Commonwealth's alleged "concession” does not change this determination, and, certainly, it does not constitute clear and convincing evidence that would rebut the presumption as to the correctness of the state court's evaluation of Dr. Wecht’s opinion.
. Evidence is not "new” if it was available at trial, but a petitioner "merely chose not to present it to the jury.” Hubbard, 378 F.3d at 340. In that situation, the choice not to present the evidence "does not open the gateway.” Id. Goldblum does not contend that Dr. Wecht was not willing and able to testify at Goldblum's trial as to his opinion. Rather, the record shows that Goldblum's attorney chose not to obtain an expert opinion for the trial.
. This dying declaration clearly has two reasonable interpretations. It is equally reasonable to believe that Wilhelm could have been saying that Miller stabbed him or that Miller, though not stabbing him, set him up for his demise.
. Goldblum also has brought to our attention a letter dated January 14, 1994, from the state trial judge supporting his application for clemency, and an affidavit dated June 17, 1998, from the trial prosecutor in which he indicates that in light of new information unavailable to him at the time of trial (he does not indicate what the new information is), he has “come to the very firm conclusion that Charles Goldblum had nothing to do with the murder of George Wilhelm other than being a frightened witness to that murder and an accessory after the fact.” While these documents are noteworthy, they are not evidence inasmuch as they merely offer personal feelings about the case presented decades after the trial and have no bearing in these proceedings given the context of this case. Rather, the two letters would be material in an attempt to obtain executive clemency, a type of proceeding not constrained by established rules of judicial procedure. Of course, we are not implying that if Goldblum seeks executive clemency or, for that matter a parole, the deciding authority should consider the letters as it is not our function to pass on that point.
.We set forth this sentence in the form of all 12 members of the jury reaching a different result because Schlup speaks in term of “no” reasonable juror voting to convict in light of the new evidence. Actually, we are confident that no reasonable juror who voted to convict would not vote to convict after considering the new evidence. Thus, by our framing of the inquiry we do not intend to imply that some jurors might change their minds but some would not.
. It is not clear to us why the experts were permitted to give their views of the significance of the dying declaration.
. Goldblum contends that "[t]he state courts were ... wrong in ruling that the expert testimony would not have led to a different result (and did not show innocence) on the theory that even if Miller was the person who stabbed Wilhelm to death, Goldblum could still have been convicted as an accomplice.” Appellant’s br. at 26. He argues that ''[i]t is impossible to determine whether the jury found [him] guilty as the stabber or as an accomplice ... [and that] it must be assumed that the jury found that he directly inflicted the mortal wounds.” Id. He builds on this contention to argue that ”[i]t is fundamental that a verdict cannot be sustained where a *230jury may have decided on one of two grounds, one of which would be impermissible," citing Yates v. United States, 354 U.S. 298, 77 S.Ct. 1064, 1 L.Ed.2d 1356 (1957), on this point.
We need not linger on this point because in United States v. Syme, 276 F.3d 131, 144 (3d Cir.2002), we indicated that we follow the rule that "[wjhen a criminal defendant appeals a conviction in which the prosecution presented more than one theory of guilt and the jury returned a general verdict ... a reviewing court should assume that the jury convicted on the factually sufficient theory and should let the jury verdict stand.” Though there is an exception in cases "if the indictment or the district court's jury instructions are based on an erroneous interpretation of law or contain a mistaken description of the law,” id. at 145, the exception is not applicable here.
In applying Syme, we point out that Dr. Wecht’s testimony merely would have been additional evidence that in no way could have undermined the Commonwealth’s accomplice theory of guilt. We will not by the use of some tortured logic confine Syme so that it does not apply here. We also observe that this case is a particularly appropriate one in which to apply Syme because even though the Goldblum jury obviously could not have considered Miller’s confession made long after Goldblum’s trial to a member of the State Board of Pardons and the attorney general of Pennsylvania, Goldblum is using the confession, which at a minimum supports an accomplice theory of guilt, as a basis for his actual innocence argument.