(dissenting).
I respectfully dissent. I agree that Tur-nage has not yet satisfied his burden of proof to establish the right to a new trial because the presentation of the handwritten recantation statement by Quantez is not sufficient to do so. But I conclude that the recantation statement is sufficient to entitle Turnage to an evidentiary hearing. I reach that conclusion by first reformulating the analytical framework used by the majority because I believe the second Larrison factor, as traditionally applied by our court, does not completely address the issues presented by the recantation statement of Quantez.
A. The Larrison Test Deals With the Withdrawal of Incriminating Testimony But Not the Addition of New Exculpatory Testimony.
All of our previous applications of the Larrison test involved trial witnesses *601whose recantations only involved the withdrawal of incriminating testimony.1 The recantation by Quantez likewise purports to withdraw Quantez’s incriminating testimony concerning Turnage — “the [testimony] given at the [trial] of Quanartis Tur-nage was lies.” But the recantation goes one step further when it also offers exculpatory testimony — “Quanartis Turnage had no knowledge of this crime other than what I told him after it became public.” Because it is undisputed that Quantez was present at the commission of the murder, his recantation statement provides new eyewitness exculpatory testimony that supports the trial testimony of Turnage that Turnage was not even present at the commission of the murder.
The traditional framing of the second Larrison factor addresses only the withdrawal of the witness’s trial testimony. It asks whether the jury might have reached a different conclusion “without the testimony.” Larrison v. United States, 24 F.2d 82, 87 (7th Cir.1928). Where, as here, the witness offers both the withdrawal of his incriminating trial testimony and the substitution of new exculpatory testimony, the traditional Larrison test does not deal with the potential impact of the new exculpatory testimony.
This appears to be a case of first impression in Minnesota because none of our former cases involved the substitution of new exculpatory testimony. There are perhaps two options to fully address this new exculpatory testimony: (1) to reformulate the second Larrison factor to ask whether the jury might have reached a different conclusion if the witness had testified consistent with his recantation statement; or (2) to bifurcate the portion of the recantation statement that purports to withdraw the witness’s incriminating trial testimony from the portion purporting to add new exculpatory testimony — dealing with the former under the Larrison test and the latter under the standard for post-conviction claims of newly discovered evidence. See Minn. R.Crim. P. 26.04, subd. 1; Sutherlin v. State, 574 N.W.2d 428, 434 (Minn.1998). The choice between these two options is important because the standard of proof required for a new trial under Larrison is lower than that required for a new trial based on newly discovered evidence.2
In choosing between these two options, we need to revisit State v. Caldwell, 322 N.W.2d 574, 584-85 (Minn.1982), where we adopted the Larrison test for recanted *602testimony. In Caldwell, we contrasted the Larrison test from that applicable to newly discovered evidence, stating:
In the case of ordinary newly-discovered evidence, as, for example, where a witness comes forward who did not testify at the defendant’s trial, a heightened standard of materiality — that the new evidence “probably would produce a different result — seems to be necessary because this new evidence has not been tried. The trial court is thus required to anticipate the effect on a second trial of evidence of unknown reliability. To avoid retrying defendants in all cases in which any new evidence, no matter how dubious or inconsequential, is discovered after trial, it became necessary to impose a rule permitting new trials only in those unusual circumstances in which the newly discovered evidence probably would produce a different result. The situation is different, however, where the evidence meets the requirements of the Larrison rule. In those eases, because the witness has recanted or other discovery has been made regarding the falsity of the testimony, the court has already been able to make a determination that the testimony actually was false; thus the matter of credibility, as it concerns a second trial, is less important. The “newly-discovered evidence” in such cases, in fact, is the absence of certain testimony, rather than the addition of new material, at a second trial. Thus the court is really considering what impact the false testimony had on the jury in the first trial; under Larri-son, it is necessary only that the jury “might have reached a different conclusion” if the testimony had not been given.
Id. at 585 (footnotes omitted).
Using this framework from Caldwell, I would conclude that the better option for dealing with exculpatory evidence that is contained in a recantation statement is to separate it from the withdrawal of the incriminating testimony and subject it to the heightened standard for newly discovered evidence. The new exculpatory testimony described in Quantez’s recantation statement, like the “ordinary newly-discovered evidence” described in Caldwell, has not been “tried” and is “of unknown reliability.” Caldwell, 322 N.W.2d at 585. The heightened standard of materiality seems appropriate to avoid automatically retrying all such cases, no matter how dubious or inconsequential the new exculpatory testimony may be.
This option is also supported by the significant retreat in the federal circuit courts from the lower standard used in Larrison. Several circuits have rejected the “might have reached a different result” standard in the second Larrison factor, and replaced it with a probability standard, at least in those cases where the government did not knowingly or negligently use false testimony. See, e.g., United States v. Williams, 233 F.3d 592, 594-95 (D.C.Cir.2000); United States v. Huddleston, 194 F.3d 214, 219 (1st Cir.1999); United States v. Sinclair, 109 F.3d 1527, 1532 (10th Cir.1997); United States v. Provost, 969 F.2d 617, 622 (8th Cir.1992); United States v. Krasny, 607 F.2d 840, 844-45 (9th Cir.1979). In fact, the seventh circuit has itself overruled this aspect of Larrison and adopted a “reasonable probability” test. United States v. Mitrione, 357 F.3d 712, 718 (7th Cir.2004).
Treating the new exculpatory testimony of Quantez as newly discovered evidence, a new trial will be granted only if it meets the four factor Rainer test:
(1) that the evidence was not known to the defendant or his/her counsel at the time of the trial;
*603(2) that the evidence could not have been discovered through due diligence before trial;
(3) that the evidence is not cumulative, impeaching, or doubtful; and
(4) that the evidence probably would produce an acquittal or a more favorable result.
Rainer v. State, 566 N.W.2d 692, 695 (Minn.1997).
Because the majority opinion only addresses the question of “whether the jury might have found the defendant not guilty if the recanting witness had not testified,” it does not resolve the newly exculpatory testimony presented in Quantez’s statement. And, although the standard for newly discovered evidence is higher than that for withdrawn testimony, the failure of Turnage to meet the second Larrison factor by the withdrawal of Quantez’s trial testimony does not necessarily compel the conclusion that Turnage could not meet the fourth Rainer factor because the focus is different — -the Larrison factor focuses on the absence of any testimony from Quantez while the Rainer factor focuses on the presence of new testimony.
If the analysis is thus reformulated into two questions, we can proceed to answer them separately.
B. Did Turnage Satisfy the Second Larrison Factor for Withdrawn Testimony?
I agree that Quantez’s recantation statement, by itself, does not sufficiently satisfy the second Larrison factor to require a new trial. But, I conclude that the statement was at least sufficient to warrant an evidentiary hearing to address each of the Larrison factors. Clearly, the postconviction court should not answer the first Lar-rison factor — whether Quantez’s trial testimony was false — without an evidentiary hearing to test that statement. We have regularly cautioned against making reliability determinations without an evidentia-ry hearing. See, e.g., Wilson v. State, 726 N.W.2d 103, 107 (Minn.2007); Opsahl v. State, 677 N.W.2d 414, 423-24 (Minn.2004).
The majority attempts to avoid this concern about the first Larrison factor by, in essence, assuming that this factor has been met and proceeding to the second Larri-son factor to determine if the absence of any testimony from Quantez might have produced a different result. To hold that the second Larrison factor was not met, the majority must conclude that the absence of testimony from Quantez could not possibly have changed the result. In other words, if it is not enough to simply conclude that the other evidence would be sufficient to convict, because we must ask the more difficult question of whether the absence of Quantez’s testimony “might” have made a difference, we should not give a negative answer to that question summarily, without a hearing, unless we are satisfied that there is no possibility that the absence could have made a difference. I find that question to be a close call. If this were the only aspect of Quantez’s recantation statement, I might be inclined to agree with the majority. But because it is not the only aspect, I respectfully disagree and conclude that an evidentiary hearing is warranted, for the reasons discussed below.
C. Did Turnage Satisfy the Fourth Rainer Factor for Newly Discovered Evidence?
Similarly, the recantation statement is not, by itself, sufficient to warrant a new trial under Rainer. But the more difficult question is whether it meets the much lower threshold to warrant an evidentiary hearing.
*604The practical consequence of the failure of the postconviction court to provide an evidentiary hearing likely is that we must either determine that the newly discovered evidence is insufficient to present genuine issues of material fact or we must assume, for purposes of a Rainer analysis, that the newly discovered evidence is true. Clearly, Quantez’s new exculpatory testimony is relevant and material and we could not conclude that the trial record “conclusively” shows that the new exculpatory testimony is false. See, e.g., Ferguson v. State, 645 N.W.2d 437, 446 (Minn.2002). And if we assume that Quantez’s new exculpatory testimony is true, the conclusion that a jury probably would not have acquitted Turnage if it had heard the exculpatory testimony is not clear, even though the Rainer standard that Turnage must meet is higher than the Larrison standard. The potential impact on the jury of the testimony of an admitted perpetrator that the defendant was not present during the commission of the crime would necessarily be greater than the impact of the failure of that perpetrator to testify at all.
The majority interprets Quantez’s recantation statement somewhat differently, concluding that it does not “include specific factual exculpatory evidence.” I agree that the recantation statement is minimal and also somewhat ambiguous. For example, when Quantez writes that his trial testimony “was lies,” it is unclear whether he is referring to all of the testimony, including his description of his role in the crime, or just his testimony implicating Turnage. But, in my view, that ambiguity underscores the need for an evidentiary hearing to sort out what Quantez currently claims is the truth. Because Quantez pleaded guilty for his involvement in the murder, and is currently serving a prison sentence under the resulting conviction, I would not read his statement to mean that he now denies his own involvement. I read his further statement, that Turnage “had no knowledge of the crime,” when made by someone who has previously admitted being present for the crime, to be a factual statement, albeit phrased in a con-clusory manner.
I recognize that the newly discovered evidence presented in Wilson v. State was stronger than that presented here — three eyewitnesses whose proposed testimony would exculpate the defendant. But I am drawn to the same conclusion that we reached in Wilson — that the newly discovered evidence was sufficient to warrant an evidentiary hearing in order to fully develop the Rainer factors, including credibility determinations. 726 N.W.2d at 107. And, because the Larrison and Rainer issues arise from the single recantation statement of Quantéz and are necessarily interrelated, I would remand to the postconviction court for an evidentiary hearing on both aspects of the recantation statement.
. See, e.g., Wilson v. State, 726 N.W.2d 103, 107 (Minn.2007) (a jailhouse informant withdrew his testimony that the defendant had confessed to committing the murder); Hooper v. State, 680 N.W.2d 89, 95 (Minn.2004) (witness recanted his trial testimony that the petitioner admitted to him that he murdered the victim); Opsahl v. State, 677 N.W.2d 414, 424 (Minn.2004) (five out of seven witnesses recanted their testimony that the defendant had made incriminating statements after the crime was committed); Dukes v. State, 660 N.W.2d 804, 809 (Minn.2003) (an eyewitness and accomplice to the murder recanted a statement that was read into evidence at the defendant’s trial, stating that the statement "was a lie”); Ferguson v. State, 645 N.W.2d 437, 446 (Minn.2002) (the recanting witness testified at trial that the defendant/petitioner told him that he was going to kill the victim and admitted to doing so shortly after the alleged murder); Flournoy v. State, 583 N.W.2d 564, 567 (Minn.1998) (the defendant's former girlfriend recanted her trial testimony that the defendant confessed to killing the victim).
. A new trial will be granted under the Larri-son test if the absence of the recanted testimony might have caused the jury to reach a different result, whereas a new trial will be granted for newly discovered evidence only if the evidence “probably would produce an acquittal or a more favorable result,” placing a significantly greater burden on the petitioner. Sutherlin, 574 N.W.2d at 434.