Ex Parte Fierro

KELLER, Judge.

OPINION

This is a post-conviction application for writ of habeas corpus filed pursuant to Article 11.07 of the Texas Code of Criminal Procedure. Applicant was convicted of capital murder and sentenced to death. We affirmed his conviction. Fierro v. State, 706 S.W.2d 310 (Tex.Crim.App.1986).

In his writ application, applicant contends that the State knowingly presented perjured testimony during a pretrial hearing on applicant’s motion to suppress his confession. Applicant claimed during the suppression hearing, and has since maintained, that his confession was coerced by the threat that his mother and stepfather were being held by the Juarez police. The El Paso police officer who took applicant’s confession, A1 Medrano, testified at the suppression hearing that he had no knowledge that applicant’s mother and stepfather were under arrest in Juarez and therefore could not have used that fact to coerce the confession. It was subsequently discovered that a supplemental offense report reflected that applicant’s parents were in the custody of the Juarez police. We accordingly abated applicant’s writ application and ordered the trial court to conduct a hearing and enter findings of fact and conclusions of law addressing applicant’s allegations and any other matters it deemed relevant to disposition of the case. Ex Parte Fierro, No. 71,899, slip op. (Tex.Crim.App. Oct. 12,1994) (unpublished).

After considering live testimony of seventeen witnesses, letter rogatory testimony,1 and numerous exhibits, affidavits and pleadings, the trial court entered findings of fact as follows:

1) That at the time of eliciting the Defendant’s confession, Det. Medrano (now deceased) did have information that the Defendant’s mother and step-father had been taken into custody by the Juarez police with the intent of holding them in order to coerce a confession from the Defendant, contrary to said Det. Medrano’s testimony at the pretrial suppression hearing.
2) That the District Attorney’s Office did not withhold this Supplemental Offense Report from the attorneys for the Defendant.
3) That Det. Medrano presented false testimony regarding the nature and extent of the cooperation between the El Paso police and the Juarez police in this particular case, as it existed in 1979. There was no evidence produced to show that such practices are still taking place.

The trial court made the following conclusions of law:

1) That there is a strong likelihood that the Defendant’s confession was coerced by the actions of the Juarez police and by the knowledge and aeqiesence [sic] of those actions by Det. Medrano.
2) It is not the conclusion of this Court that the Defendant should be released from these charges, but that he should be retried by another jury who will then render a verdict based on all the evidence, both old and new that has been developed at these hearings.

Applicant urges that we adopt the findings of the trial court and grant him relief based upon his claim that the State used perjured testimony. After reviewing the record, we find that the trial court’s factual findings are adequately supported. As a result of the trial court’s findings and *372the evidence in the record, we conclude that applicant’s due process rights were violated by Medrano’s perjured testimony.2 But, because we conclude that the error was harmless, we deny relief.3

Traditionally, on habeas corpus review involving federal constitutional error that is subject to a harmless error analysis, we have placed on the defendant the burden of proving, by a preponderance of the evidence, that the eiTor contributed to his conviction or punishment. Ex Parte Dutchover, 779 S.W.2d 76, 78 (Tex.Crim.App.1989). See also Ex Parte Crispen, 777 S.W.2d 103, 109 n. 6 (Tex.Crim.App.1989) (plurality opinion) (Clinton, J. concurring). Ex Parte Barber, 879 S.W.2d 889, 891-892 (plurality opinion) & 893 (Meyers, J. concurring) (Tex.Crim.App.1994), cert. denied, — U.S.-, 115 S.Ct. 739, 130 L.Ed.2d 641 (1995). We apparently made an exception in the case of “knowing use of perjured testimony” by holding that, when such is shown, the Rule 81(b)(2) harmless error standard applies.4 Adams, 768 S.W.2d at 292. Castellano, 863 S.W.2d at 485. In applying Rule 81(b)(2), we relied upon Napue v. Illinois, 360 U.S. 264, 79 S.Ct. 1173, 3 L.Ed.2d 1217 (1959) and United States v. Bagley, 473 U.S. 667, 105 S.Ct. 3375, 87 L.Ed.2d 481 (1985), which held that error in the admission of perjured testimony was governed by a “materiality” standard identical to the harmless error standard announced in Chapman v. California, 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967), upon which Rule 81(b)(2) was based. Adams, 768 S.W.2d at 292. But Napue and Bagley both involved direct rather than collateral attacks. The Supreme Court has now clarified that the Chapman standard need not apply to collateral review of federal constitutional trial errors. Brecht v. Abrahamson, 507 U.S. 619, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993).5

While some constitutional errors are not subject to a harmless error analysis, “they are the exception and not the rule.” Rose v. Clark, 478 U.S. 570, 578, 106 S.Ct. 3101, 3106, 92 L.Ed.2d 460 (1986). “[I]f the defendant had counsel and was tried by an impartial adjudicator, there is a strong presumption that any other errors that may have occurred are subject to a harmless error analysis.” Id. at 579, 106 S.Ct. at 3106-07. An error is subject to a harmless error analysis unless it is “structural.” Arizona v. Fulminante, 499 U.S. 279, 309, 111 S.Ct. 1246, 1264, 113 L.Ed.2d 302 (1991). An error is “structural” only if it is the kind of error that affects the framework in which the trial takes place and defies analysis by harmless error standards. Id. at 309-310, 111 S.Ct. at 1264-65. The Supreme Court has recognized the following as examples of structural error: total deprivation of counsel at trial, a biased judge, the unlawful exclusion of members of the defendant’s race from a grand jury, denial of the right to self-representation at trial, and denial of the right to a public trial. Id. On the other hand, “trial” error is error which occurs during the presentation of evidence at trial, and may “be quantitatively *373assessed in the context of other evidence presented in order to determine whether its admission was harmless.” Id. at 307-308, 111 S.Ct. at 1263-64.6

*372If the appellate record in a criminal case reveals error in the proceedings below, the appellate court shall reverse the judgment under review, unless the appellate court determines beyond a reasonable doubt that the error made no contribution to the conviction or punishment.

*373The “knowing use of perjured testimony” is clearly error involving the presentation of evidence, and as such, carries all the hallmarks of trial error. The error is quite unlike the examples of structural error listed by the Supreme Court. Those examples defy quantitative comparison because the entire trial process is tainted by the constitutional violation. Not so with perjured testimony: ordinarily, there will also be non-perjured evidence, which would not carry the taint of the constitutional violation. Whether the perjured testimony harmed the defendant can be quantitatively assessed by examining the remaining evidence at trial and the effect of the perjured testimony upon that evidence.

Moreover, the fact that Bagley contains a materiality standard is an indication that Bagley error is not structural. This conclusion is buttressed by the fact that the Supreme Court has held the materiality standard for the knowing use of perjured testimony to be equivalent to the Chapman harmless error standard. See Bagley, 473 U.S. at 679 n. 9, 105 S.Ct. at 3382 n. 9. Moreover, the Supreme Court has held that the Bagley standard may be legitimately characterized as either a “harmless error” standard or a “materiality” standard: “Although this rule is stated in terms that treat the knowing use of perjured testimony as error subject to a harmless error review, it may as easily be stated as a materiality standard.” Id. at 679-680, 105 S.Ct. at 3382. The Supreme Court made clear that, unlike error arising from admission of perjured testimony, error in the suppression of exculpatory evidence would be subject to a less stringent harmless error standard on direct appeal than that ordinarily accorded to federal constitutional trial errors.7 473 U.S. at 678-681, 105 S.Ct. at 3381-83. Hence, the Supreme Court appears to have at least implicitly held that the knowing use of perjured testimony is the kind of error that is subject to a harmless error analysis.

Because the materiality standard for perjured testimony is identical to the Chapman harmless error standard, a finding of materiality would obviate the need to conduct a harmless error analysis on direct appeal. The Supreme Court has not expressly addressed the issue of whether a separate harmless error standard applies on collateral review to the knowing use of perjured testimony. The Court has addressed the applicability of a separate harmless error standard to the suppression of exculpatory evidence. It rejected the standard in that context because the materiality standard for such error is already more lenient (to the State) than the federal habeas harmless error standard. Kyles v. Whitley, 514 U.S. -, - and -n. 9, 115 S.Ct. 1555, 1567 and 1567 n. 9, 131 L.Ed.2d 490, 507 and 507 n. 9 (1995).8 But, because the materiality standard for the *374knowing use of perjured testimony is the Chapman harmless error standard, the materiality standard is more stringent (on the State) than either the state or federal habeas harmless error standards. This leaves open the possibility of applying a separate harmless error standard on collateral review.

In fact, in the wake of Brecht (holding Chapman to be the inappropriate harmless error standard for federal habeas), two federal circuits have expressly held that the federal habeas harmless error standard applies to error based upon the knowing use of perjured testimony when that error is raised on collateral review. Gilday v. Callahan, 59 F.3d 257, 268 (1st Cir.1995), cert. denied, — U.S.-, 116 S.Ct. 1269, 134 L.Ed.2d 216 (1996). Robinson v. Arvonio, 27 F.3d 877, 884-886 (3rd Cir.1994), vacated for farther consideration in light of O’Neal v. McAninch, — U.S. -, 115 S.Ct. 1247, 131 L.Ed.2d 129 (1995).9 The Third Circuit discussed the distinction between structural error and trial error and held that the knowing use of perjured testimony was trial error. Robinson, 27 F.3d at 883-884. The First Circuit noted that the materiality standard for the knowing use of perjured testimony is “friendlier” to the defendant than the habeas harmless error standard, and therefore, both standards had to be satisfied: the materiality standard to establish the due process violation and the habeas standard to establish harm on collateral review. Gilday, 59 F.3d at 268. In accordance with the above discussion, we hold that the knowing use of perjured testimony is trial error, subject to the harmless error standard applicable on habeas corpus.

Moreover, there is another compelling reason to apply a harmless error analysis to the present case. The perjured testimony here was not used as substantive evidence. Instead, the testimony in question related solely to the admissibility of a confession. The only harm that could possibly flow from the testimony is the admission into evidence of an inadmissible involuntary confession. The Supreme Court has clearly held that the improper admission of an involuntary confession is “trial” error, subject to a harmless error analysis. Fulminante, 499 U.S. at 306-312, 111 S.Ct. at 1262-1266. In fact, the Supreme Court characterized it as a “classic ‘trial error’ ” that “is markedly different” from the constitutional violations held to be structural. Id. at 309, 111 S.Ct. at 1264. If the admission of an involuntary confession is subject to a harmless error analysis, then how can error based upon the knowing use of perjured testimony, the only consequence of which may have been the admission of an involuntary confession, not be subject to a harmless error analysis? Obviously, if the improper admission of a confession is subject to harmless error analysis, then the knowing use of perjured testimony to procure the admission of that confession must likewise be subject to harmless error analysis. If the admission of the confession were in fact harmless, it would make no sense to grant relief merely because the admission of that confession was procured through perjured testimony in a pretrial hearing.

Because the error in the present ease is trial error rather than structural error, application of the habeas harmless error standard is appropriate.10 Therefore, we *375hold that applicant has the burden to prove by a preponderance of the evidence that the error contributed to his conviction or punishment.11

In addition to the confession, the State also utilized the testimony of Gerardo Olague, an eyewitness to the murder. While applicant’s confession alleged that Olague was an accomplice, there is no other evidence in the record to that effect.12 Thus, had the confession been excluded, there would have been no need to corroborate Olague’s testimony. Olague testified in great detail to witnessing applicant commit the murder. In the absence of applicant’s confession, there would be no evidence to contradict Olague’s statements of how the offense happened. Other than the accomplice allegation, applicant has shown no reason to doubt Olague’s testimony.13

*376The difference between the Chapman and the habeas harmless error standards is the difference between a possibility and a probability. See Gilday, 59 F.3d at 269. We agree that the applicant has met the Chapman standard. The error likely caused the admission into evidence of an involuntary confession, thus raising the possibility that the error contributed to the verdict. But, given Olague’s eye-witness testimony and the lack of any real reason to doubt his credibility, it is more probable than not that the outcome of applicant’s trial would have been the same absent the confession.14 Under these circumstances, the ap*377plicant has failed to carry his burden of proving harm by a preponderance of the evidence.15

Accordingly, the application for writ of ha-beas corpus is DENIED.

. A letter rogatory is described as:

A formal communication in writing, sent by a court in which an action is pending to a court or judge of a foreign country, requesting that the testimony of a witness resident within the jurisdiction of the latter court may be there formally taken under its direction and transmitted to the first court for use in the pending action.

Black's Law Dictionary 471 (5th ed. 1983); see Fed.R.Civ.P. 28.

. Due process prohibits prosecutors from presenting testimony that any member of the "prosecution team,” including both investigative and prosecutorial personnel, knows to be false. Ex Parte Adams, 768 S.W.2d 281, 292 (Tex.Crim.App.1989). Ex Parte Castellano, 863 S.W.2d 476, 485 (Tex.Crim.App.1993).

. Applicant also contends that the State suppressed the supplemental offense report in violation of Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963). Assuming, without deciding, that applicant is correct, we note that the only result of such suppression is that the peijured testimony remained unrevealed. Because prosecutors have a duty to reveal the perjured testimony of members of the "prosecution team,” whether or not the prosecutor has knowledge of such perjury, this claim effectively merges into the perjured testimony claim and is properly analyzed under the standards pertaining to the latter.

. Tex.R.App.P. 81(b)(2) states:

. In the federal system, a constitutional error is harmful on collateral review only if it had a "substantial and injurious effect or influence in determining the jury's verdict.” Brecht, 507 U.S. at 637, 113 S.Ct. at 1722, 123 L.Ed.2d at 373. The State bears the burden of proving an error to be harmless. See O'Neal v. McAninch, 513 U.S. 432, 115 S.Ct. 992, 130 L.Ed.2d 947 (1995).

.Judge Maloney's dissenting opinion argues that footnote 9 of Brecht exempts some "especially egregious” trial errors from a harmless error analysis. But, our reading of the footnote merely indicates that the Court reserved that issue for another day:

Our holding does not foreclose the possibility that in an unusual case, a deliberate and especially egregious error of the trial type, or one that is combined with a pattern of prosecutorial misconduct, might so infect the integrity of the proceeding as to warrant the grant of habe-as relief, even if it did not substantially influence the jury's verdict. [Citation to concurring opinion omitted]. We of course are not presented with such a situation here.

507 U.S. at 637 n.9, 113 S.Ct. at 1722 n. 9, 123 L.Ed.2d at 373 n. 9 (emphasis added). Moreover, even if Brecht had definitively held that some "especially egregious” errors were immune to a harmless error analysis, such a conclusion would not mean that the error in this case was such an error. Given the strong presumption that federal constitutional trial errors are subject to harmless error standards, a particular type of trial error should be held subject to a harmless error analysis in the absence of clear direction to the contrary from the Supreme Court.

. The standard is the one applied in Strickland v. Washington, 466 U.S. 668, 694, 104 S.Ct. 2052, 2068, 80 L.Ed.2d 674 (1984) which places the burden on the defendant to show that “there is a reasonable probability that, [but for the error] ... the result of the proceeding would have been different.” Bagley, 473 U.S. at 682, 105 S.Ct. at 3383 (ellipse and bracketed material added).

. Compare footnotes 5 and 7, infra.

. The Third Circuit erroneously held that Brecht shifted the burden of proof upon the petitioner. See Robinson, 27 F.3d at 885. The Supreme Court's action in vacating the decision merely addresses the nature of the habeas standard. It does not undermine the Third Circuit's holding that the habeas standard (whatever it is) applies to the error involving the knowing use of perjured testimony.

. Judge Clinton's dissenting opinion argues that a separate, habeas harmless error analysis should not apply because applicant "could not have raised” the claim, "even by due diligence, on direct appeal." However, contrary to Judge Clinton's opinion, the record shows that applicant had the opportunity to discover the basis of his claim in time to advance it at trial or in a motion for new trial. The perjured testimony occurred at a pretrial suppression hearing more than two months before trial. Because Medra-no's perjured testimony related to statements he made to applicant, applicant was obviously aware, and in fact contended, at the pretrial hearing, that Medrano’s testimony was untrue.

Moreover, the supplemental offense report— which contained in its second paragraph the facts revealing Medrano’s perjury — was available at the suppression hearing. Testimony at the *375habeas hearing showed that defense counsel looked at the supplemental offense report during the suppression hearing and cross-examined a witness concerning the report. Evidence at the habeas hearing also supports the State's contention that the report was contained in the State's file before and during trial and that the State’s file was open to the defense for examination. The evidence also supports the conclusion that defense counsel did not read the report’s second paragraph. The habeas trial court apparently believed this evidence because it found that the State did not withhold the supplemental offense report. Hence, the facts supporting applicant’s claim were available in time to raise the issue on direct appeal.

We express no opinion regarding whether an applicant may avail himself of the Chapman harmless error standard if he can demonstrate that he had no opportunity to raise the issue on direct appeal. In the present case, applicant has not demonstrated that he lacked such an opportunity.

. As noted in footnote 5, the federal habeas standard differs from our formulation of the state standard. But, because we are not bound to apply federal harm standards to our post-conviction writs, we apply the state standard announced in Dutchover.

Judge Clinton accuses us of impliedly adopting the Brecht ¡O'Neal federal habeas standard and then "distorting” it. He claims that we distort that standard by using the word “contribution’’ and by adopting the preponderance level of confidence. But in his own concurring opinion in Crispen, Judge Clinton stated:
In a collateral attack, however, where the burden is upon the applicant to establish the illegality of his constraint, we might justifiably require an applicant to plead and prove that asserted constitutional error in fact more likely than not contributed to his conviction.

777 S.W.2d at 109 n. 6. (Emphasis added.) It is clear that we adopted this very portion of his opinion as the law of Texas in Dutchover. 779 S.W.2d at 78.

In his dissent, Judge Baird claims that the Chapman standard should be used on habeas because Adams and Castellano, two habeas cases, used that standard. He argues that our opinion suggests that the Adams and Castellano Courts were unaware that the proceedings involved collateral attacks. This opinion makes no such sug-gestión. We do submit that the Courts in those cases were unaware of the significance of such proceedings on a harmless error analysis. Dutchover, which formulated the general harmless error test for state habeas, was decided eight months after Adams. Although decided later, Castellano does not make a single reference to Dutchover, or even to Brecht. The only habeas case that Castellano cites in support of a Chapman harmless error analysis is Adams. Neither Adams nor Castellano discussed the potential significance of a case's posture as a collateral attack on the materiality/harm determination. Moreover, the federal decisions applying a separate habeas harm standard to perjured testimony were decided after both Adams and Castellano.

Judge Baird further claims that we “do not explain why or how” application of the Chapman standard is erroneous in the present context. He has apparently missed the point of our entire opinion. Our thesis is simple: the knowing use of perjured testimony is the type of error that is subject to a habeas harm standard, and the appropriate standard of harm on habeas review is the one announced in Dutchover rather than the Chapman standard. We think the supporting reasoning for this thesis is made clear enough in our opinion.

. When questioned about the contents of Ola-gue’s written statement, Detective Medrano testified that Olague said that "they both dragged the body.” An examination of Olague’s statement shows that Medrano’s testimony in this respect was inaccurate. Given the nature of this testimony, we believe it does not raise the issue of accomplice status.

. For the most part, we disagree with Judge Maloney’s characterization of the evidence regarding Olague’s credibility. For example, Judge Maloney claims there was some indication that a deal had been reached between Olague’s attorney and the prosecution. But, although defense counsel asked Olague whether he knew of such a deal, Olague said he did not, and the defense never elicited any evidence that there was a deal. More important, Olague was not even appointed an attorney until after he had told the police what occurred during the offense. Judge Maloney is also skeptical about Olague’s delay in going to the police. But, the mere fact *376that he came forward, when the police were not even aware of his existence, weighs in Olague’s favor. Moreover, Olague’s written statement explains that he was afraid applicant would kill him. In the end, he revealed what had happened even though he believed “it could mean [his] life if Cesar Fierro found out.’’ And finally, Judge Maloney finds a discrepancy in Olague's testimony about how long he had known applicant. Although there is testimony that could be described as ambiguous regarding when Olague met applicant, in our view, Olague's testimony is consistent in that regard — testimony about dealing with applicant for several months refers to the months before Olague went to the police, not the months before the murder. When Olague’s testimony is read in its entirety, and viewed in light of the rest of the evidence, his credibility is solid.

. Judge Clinton argues that we erroneously impose an outcome-determinative test for harm. He supports his argument by claiming that neither the Chapman nor the Brecht tests are outcome-determinative, and he claims that the Supreme Court manifested its intent with regard to those tests by clearly formulating an outcome-determinative test in Strickland. Of course, Dutchover employs neither the Chapman nor the Brecht tests for harm. Even assuming some relationship to those federal harmless error tests (the word "contribution" is, after all, derived from Chapman), we reject Judge Clinton’s premise.

Judge Clinton apparently views the harmless error standards as containing three components: (1) burden of proof, (2) nature of the harm, and (3) level of confidence that the harm will occur. In our view, Judge Clinton’s supposed prongs two and three are actually a single inquiry: the level of confidence to which a reviewing court believes the error affected the outcome. An evaluation of Supreme Court pronouncements concerning the Chapman and Brecht standards supports this conclusion. Several times, the Supreme Court has expressly referred to the Chapman standard as determining the effect of an error on the “outcome” of the case. Satterwhite v. Texas, 486 U.S. 249, 256, 108 S.Ct. 1792, 1797, 100 L.Ed.2d 284 (1988). Berkemer v. McCarty, 468 U.S. 420, 444-445, 104 S.Ct. 3138, 3152-53, 82 L.Ed.2d 317 (1984). This issue was addressed in detail in Rose v. Clark, 478 U.S. 570, 106 S.Ct. 3101, 92 L.Ed.2d 460 (1986):

Harmless error analysis addresses a different question: what is to be done about a trial error that, in theory, may have altered the basis on which the jmy decided the case, but in practice clearly had no effect on the outcome? This question applies not merely to Sandstrom violations, but to other errors that may have affected either the instructions the jury heard or the record it considered — including errors such as mistaken admission of evidence, or unconstitutional comment on a defendant’s silence, or erroneous limitation of a defendant’s cross-examination of a prosecution witness. All of these errors alter the terms under which the jury considered the defendant’s guilt or innocence, and therefore all theoretically impair the defendant’s interest in having a jury decide his case. The dissent’s argument — that the Sixth Amendment forbids a reviewing court to decide the impact of a trial error on the outcome [citation omitted] logically implies that all such errors are immune from harmless-error analysis. Yet this Court repeatedly has held to the contrary.

Id. at 582 n. 11, 106 S.Ct. at 3108 n. 11. The Supreme Court has further referenced the Chapman standard as determining whether error is “harmless beyond a reasonable doubt.” Kyles, -U.S. at-, 115 S.Ct. at 1567, 131 L.Ed.2d at 507. Use of the general word "harmless” in connection with a level of confidence indicates that the nature of the harm is the same for various harmless error standards — whether the error affected the outcome — and the difference between the standards centers on the burden of proof and the level of confidence.

Likewise, cases involving the Brecht standard show that it is also outcome-determinative. Brecht adopted, for collateral review, the harmless error standard announced in Kotteakos v. United States, 328 U.S. 750, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946). Brecht, 507 U.S. at 621, 113 S.Ct. at 1713, 123 L.Ed.2d at 363. Kotteakos recognized that harmlessness is determined by taking into account the error’s likely effect on the outcome of the trial. Id-, at 764, 66 S.Ct. at 1247. In United States v. Lane, 474 U.S. 438, 106 S.Ct. 725, 88 L.Ed.2d 814 (1986), the Supreme Court held that an error was harmless under Kotteakos because of the "overwhelming evidence of guilt.” Lane, 474 U.S. at 450, 106 S.Ct. at 732. The Court explained that the Kotteakos standard was not a sufficiency of the evidence inquiry but differed in the level of confidence needed to establish harm:

[T]he harmless error inquiry is entirely distinct from a sufficiency-of-the-evidence inquiry.... But that does not mean that overwhelming evidence of guilt is irrelevant; the threshold of overwhelming evidence is far higher than mere sufficiency to uphold conviction.
*377Lane, 474 U.S. at 450 n. 13, 106 S.Ct. at 732 n. 13. Recently, the Court reiterated that, under Kotteakos, "the error must have been prejudicial: It must have affected the outcome” of the proceedings. United States v. Olano, 507 U.S. 725, 733, 113 S.Ct. 1770, 1777, 123 L.Ed.2d 508, 519 (1993).
We agree with Judge Clinton that "the difference in language” between the various harmless error standards "stands for something.” The difference in language signifies different degrees of confidence. The phrase "substantially and injuriously affect and influence the jury's verdict" in Brecht simply reflects a greater degree of confidence in the effect on the outcome than the phrase "beyond a reasonable doubt did not contribute to conviction or punishment” found in Chapman.

. All four dissenting opinions argue that the prosecutor's affidavit demonstrates harm by alleging that he would have dismissed the case unless he could have corroborated Olague’s testimony. While the prosecutor’s view of the evidence at trial is itself some evidence to consider in determining whether applicant has met his burden of proof on habeas corpus, it is not controlling. For the reasons stated in the body of this opinion, we believe that the evidence at trial reveals that Olague’s testimony was credible and did not need to be corroborated. Moreover, much of the physical evidence was consistent with Olague’s testimony and his statements to the police — corroborating Olague’s story. While such evidence may not be the kind of corroboration envisioned by the accomplice witness rule, as we have already noted, without the confession, there was no evidence that Olague was an accomplice. Ironically, this fact was made very clear by the prosecutor in his closing argument to the jury. Further, the source of mistrust for accomplice testimony is the possibility that the accomplice is attempting to shift blame for a crime in which he was involved. But, the defense never made such a claim. Instead, the defense theory in closing argument centered on whether Olague was present when the murder occurred. The defense argued that Olague was not present but made up the entire story. The consistency between Olague's testimony, his statements to the police, and the physical evidence was more than enough corroboration to establish Olague's credibility on that matter.

Judge Clinton’s contention that we must look at the trial that occurred (i.e. there was evidence of accomplice status because the confession was admitted) appears to be an attempt to have one’s cake and eat it too. In evaluating harm. Judge Clinton wishes to consider parts of the confession benefitting the State but wants to ignore the parts that benefit appellant. We believe that approach would distort the harmless error inquiry-

In analyzing harm, Judge Baird focuses on the motion to suppress and appeal stages of the case but neglects the trial stage. Obviously, an error occurring within a motion to suppress proceeding must have an adverse effect on the outcome of that proceeding in order for harm to exist. But, an exclusive focus on the motion to suppress proceeding does not go far enough. If the erroneous admission of the contested evidence (i.e. the confession) did not affect the outcome of the trial, then the error is harmless, despite the fact that it affected the outcome of the suppression proceeding, resulting in the admission of the evidence at trial. As for appeal, the error in the present case occurred during trial, not in connection with appeal. Judge Baird’s attempt to focus on the outcome of an appeal would collapse all habeas cases to direct appeal standards because the outcome of an appeal would necessarily turn on the application of a direct appeal standard of harm. Having decided that a different standard of harm applies to habeas corpus, we reject an indirect attempt to collapse those standards.

Finally, Judge Baird criticizes our opinion for converting the materiality requirement into a pure harm analysis. But, even the other dissenting opinions recognize the overlap between materiality and harm concepts, and Supreme Court caselaw supports treating the concepts the same. Nevertheless, there is an analytical difference that our opinion continues to respect: materiality is a component of the due process violation, and hence a component of the error, while harm is a separate inquiry. Viewed from that perspective, the materiality standard for perjured testimony remains what it has always been — the Chapman test. But the harm standard on collateral review is the Dutchover test. When the materiality standard is the same as or more lenient (to the State) than the harm standard, then a harm analysis is of course redundant. Establishing the error would automatically establish harm, and so it is on direct appeal. But, in the case of perjured testimony, the materiality standard is less lenient (to the State) than the habeas harm standard. Thus, establishing error does not automatically establish harm on habeas corpus, and harm must be shown in accordance with Dutchover.