Gordon McCloud, J.
¶1 — Julio Davila challenges the Court of Appeals’ decision affirming his conviction for second degree murder. At issue is whether the State violated its disclosure obligation under Brady v. Maryland, 373 U.S. 83, 83 S. Ct. 1194, 10 L. Ed. 2d 215 (1963), when it failed to disclose the fact that a forensic scientist who analyzed certain DNA (deoxyribonucleic acid) evidence used in Davila’s case had been fired for incompetence. While we conclude that this evidence was both favorable to the defendant and suppressed by the State in violation of its Brady obligations, we also conclude that the evidence was not material to the outcome of Davila’s trial, given the specific facts presented in this case. We therefore affirm.
FACTS
¶2 In 2007, Jeramie Davis called 911 to report an assault at an adult bookstore in Spokane. At the store, police found the owner, John Allen, lying unconscious on top of a baseball bat and bleeding from his head. Allen later died of his injuries.
¶3 While police were processing the crime scene, some of Allen’s relatives arrived and told them that Allen’s car was missing from its usual parking place outside the store. Police found the car a short time later, less than a block from the store, with its passenger side door slightly ajar.
¶4 In 2008, Davis was convicted of Allen’s murder based on evidence that he had robbed Allen’s store on the night that Allen was killed. Davila had not yet become a suspect.
¶5 Davila would, however, become a suspect later. In the investigation leading to Davis’ trial, detectives swabbed [59]*59four areas of the baseball bat found at the scene and several areas of Allen’s car in order to test them for DNA evidence. These swabs were later tested by former Washington State Patrol Crime Laboratory (Crime Lab) forensic scientist Denise Olson.
¶6 At issue in this case are a swab taken from the handle of the baseball bat (Swab D) and a swab from the steering wheel of the car (Item 24). Olson created a profile for Swab D that revealed a mixture of at least two different people’s DNA. Olson found that the “major [DNA] contributor” was “Unknown Individual A.” Clerk’s Papers (CP) at 275 (emphasis omitted). She found that the other contributor might be Allen but was definitely not Davis. Olson also created a profile for Item 24. She found that Allen was included as a DNA contributor, “Unknown Individual A” could not be excluded as a contributor, and Davis was excluded. CP at 279 (emphasis added and omitted). In other words, Olson’s analysis of Item 24 was inconclusive as to whether that item contained DNA from “Unknown Individual A.” Id. (emphasis omitted).
¶7 The DNA profile of “Unknown Individual A” (from Swab D) was entered into the Combined DNA Index System (CODIS) database. 3 Verbatim Report of Proceedings (VRP) at 434-35 (July 12, 2012). In 2011, three years after Davis’ conviction, the Crime Lab received a “hit, or a match in that database” between the DNA from “Unknown Individual A” and DNA from the defendant in this case, Julio Davila. Id. at 435.
18 Lorraine Heath, the supervising forensic scientist at the Crime Lab, retested and analyzed Swab D (the baseball bat swab) and compared it against a new reference swab obtained from Davila. Her testing confirmed that the DNA from Swab D matched the DNA from the reference swab. Heath also retested Item 24 (the steering wheel swab) and confirmed that Davila could be neither included nor excluded as a contributor of DNA on that sample. In other words, Heath’s retesting confirmed Olson’s results: Swab D [60]*60contained DNA from “Unknown Individual A,” later identified as Davila’s DNA, and Item 24 was inconclusive for this DNA.
¶9 The State charged Davila with Allen’s murder under two different theories: (1) first degree felony murder for working with Davis to commit a robbery and thereby causing Allen’s death and (2) second degree felony murder for causing Allen’s death in the course of an assault or attempted assault in the second degree. The first degree felony murder charge was predicated on the theory that Davila worked with Davis to rob Allen’s store, but the court dismissed that charge before the jury deliberated, finding that there was insufficient evidence that Davila and Davis knew one another.
¶10 At trial, the State presented the following evidence: (1) fingerprints taken from a glass counter close to where Allen was found matched Davila’s, (2) Davila’s DNA was found on the handle of the baseball bat used to murder Allen, (3) Davila claimed never to have been in Allen’s store or to have had any contact with Allen, and (4) Davila lived behind Allen’s store at the time of the murder.
¶11 The jury convicted Davila of second degree murder. It also found that Davila was armed with a deadly weapon at the time. On October 25, 2012, the court sentenced Davila to 199 months in total confinement, including 24 months for the deadly weapon enhancement.
PROCEDURAL HISTORY
¶12 On July 25,2012, after Davila’s conviction but before his sentencing, the defense filed a motion for a new trial. CP at 162-65. It alleged prosecutorial misconduct in closing argument and rebuttal, and it also asserted that the State withheld Brady material: the fact that “a forensic expert closely linked to the case was incapable of doing her job” and that she had been fired for incompetence. CP at 162.
¶13 The expert in question was Olson, the scientist who first tested Swab D—the baseball bat swab that eventually [61]*61registered the “hit” in the CODIS database—and Item 24—the steering wheel swab that was inconclusive for Davila’s DNA. After filing a public disclosure request, the defense learned, after the guilty verdict but prior to sentencing, that Olson had been fired in 2011 after receiving poor evaluations for roughly five years. It also learned that the Crime Lab had performed an audit of Olson’s work in 2007, the year that she tested Swab D and Item 24 for Davis’ case, and that this audit had revealed errors in the vast majority of Olson’s cases and had “resulted in ‘Brady letters’ being sent to eleven prosecuting attorneys notifying them of [Olson’s] problems and her faulty results.” CP at 256.
¶14 In response to the motion for a new trial, the State argued that Olson had not performed any of the DNA testing crucial to the case against Davila. Specifically, it asserted that Heath had conducted the DNA tests in Davila’s case and had also “reviewed all of Ms. Olson’s tests from the Davis case and agreed with all of Ms. Olson’s findings and conclusions.” CP at 261. The State also asserted that when Heath matched Davila’s DNA to the DNA found on the handle of the murder weapon, she did not rely on any of the testing done by Olson. CP at 267.
¶15 Finally, the State did not dispute the fact that it never disclosed this information—Olson’s substandard work performance and the pretrial audits during which the State documented it—to the defense. Instead, the State argued that defense counsel knew at the time of jury selection that Olson no longer worked in the Crime Lab, so he could have discovered the reason with due diligence. CP at 268.
¶16 The trial court ultimately held three separate hearings on Davila’s motion for a new trial, granting the defense two extensions of time to obtain evidence relevant to the Brady claim.
[62]*62
The First Hearing on the Motion for a New Trial
¶17 The trial court first addressed the defense’s motion for a new trial on August 1, 2012, the date set for Davila’s sentencing. The State acknowledged that Olson’s performance record “standing alone . . . would cause a great deal of concern.” 4 VRP at 583 (Aug. 1, 2012). But it argued that in light of the fact that Heath retested Swab D and Item 24 and confirmed Olson’s results, there was no reasonable probability that the disclosure of Olson’s performance record and termination would have changed the outcome of Davila’s trial. In making this argument, the prosecutor cited Kyles v. Whitley, 514 U.S. 419, 115 S. Ct. 1555, 131 L. Ed. 2d 490 (1995), and United States v. Bagley, 473 U.S. 667, 105 S. Ct. 3375, 87 L. Ed. 2d 481 (1985), the key United States Supreme Court cases defining Brady materiality.
¶18 The court also stated that it had “no evidence to find that [the prosecutor] withheld the . . . report [on Olson’s incompetence]” and that the prosecutor “had no obligation to find out about the extent of the report.” 4 VRP at 596 (Aug. 1, 2012). Nevertheless, it concluded that Heath’s retesting would be grounds for a new trial if it were shown that Olson could have contaminated the samples that Heath later retested. Citing In re Personal Restraint of Stenson, 174 Wn.2d 474, 276 P.3d 286 (2012), a Brady case involving mishandled forensic evidence, the trial court ruled that the defense was entitled to an evidentiary hearing on whether any of Olson’s initial testing could have compromised the DNA evidence in Davila’s case. Specifically, the trial court ruled that “the issue is contamination [,] ... meaning] the material was compromised, and no matter how many times you test it, it will not make any difference. That is what the defense has to show for the court to make any changes.” 4 VRP at 598-99 (Aug. 1, 2012); CP at 295-96.
¶19 The defense suggested that it would need an expert to determine whether the DNA evidence used against Da-[63]*63vila might have been compromised,1 but the trial court did not limit the subject of the evidentiary hearing it was granting to expert testimony. It made clear that there was a factual issue as to whether the DNA evidence used to convict Davila might have been mishandled and that defense counsel was entitled to explore that issue. Indeed, the court suggested that there would be no need to consult an expert until the defense had investigated the facts surrounding the Crime Lab’s handling of the DNA evidence: “I am willing ... to give you an opportunity to basically demonstrate that you have sufficient evidence . . . [that] this was contaminated or these procedures were not followed and it was likely this was contaminated. ... I would give you the authority to depose .... Then if you are going to need an expert after you have deposed, come to me so we can have one hearing.” 4 VRP at 597-98 (Aug. 1, 2012) (emphasis added). The court even called the parties’ attention to a recent evidentiary hearing it had just conducted in another case, to give the parties an example of how they might prepare for such a hearing. Id. at 598 (“If you took a look at that fairly extensive affidavit that is in State v. Woods, it might be of assistance to both counsel in looking at when somebody is arguing a contamination theory.”). And it granted a continuance of several weeks for the defense to conduct this factual investigation. Id. at 600-02 (scheduling September 7 status conference on motion for new trial).
The Second Hearing on the Motion for a New Trial
¶20 The court reconvened about two and a half months later, on October 19, 2012, but the defense had not obtained any evidence of contamination. Defense counsel explained that he had contacted a DNA expert but needed more time to explore the possibility that the DNA evidence used in [64]*64Davila’s case had been contaminated. He stated that the expert had determined that contamination might have occurred in either of two ways: (1) if “Mr. Davila’s DNA had been at the lab prior to the evidence from this case arriving at the lab” or (2) “cross contamination of evidentiary items.” 4 VRP at 605 (Oct. 19, 2012). But he asked for more time to explore those possibilities and suggested that Davila’s DNA might have entered the Crime Lab when detectives investigated prior burglaries of Allen’s bookstore. Id. at 606.
¶21 The trial court expressed frustration at defense counsel’s failure to provide more than speculation about the presence of Davila’s DNA in the Crime Lab:
I would expect that you would have done some investigation with regard to the speculation about what happened on these robberies [sic]. Did, in fact, you talk to the detectives or the police officers? Did you see what, in fact, was sent to the crime lab? Your expert’s not going to know that or have any better way of finding that out than you guys are. I would expect to see some sort of affidavit or declaration saying “here is what we have learned from talking with the investigators, from talking with the lab.” Not about contamination, but just was Mr. Davila’s DNA there?
... So the question becomes what . . . was in the lab that might have had Mr. Davila’s DNA on it in order to contaminate it? And that is a question that I do not need a DNA expert to answer . . . because it is a factual question. Whether or not it was contaminated, that is what the DNA expert needs to answer for me. But before that expert can give me an answer, I presume he or she needs to know whether or not Mr. Davila’s DNA was in the lab, on something that was in the lab at the same time as the bat was tested. Then you turn it over to the expert and say[/]ask what they think about it. Could it be contaminated or not? But if we do not have any evidence that Mr. Davila’s DNA was in the lab at the time that the bat was tested, how can we have contamination if it was not there?
Id. at 609-11. Defense counsel agreed that “we have to sort of pin point a situation where we can say, more than likely, that could have been Mr. Davila’s DNA, that may have gone [65]*65over on a piece of evidence, and .. . Ms. Olson touched that piece of evidence.” Id. at 612.
¶22 The trial court determined that it could not consider the motion for a new trial until it had evidence, “on a more likely than not basis, [that] Mr. Davila’s DNA was on something that was in the lab at the relevant time.” Id. at 612-13. It gave defense counsel until later that same week to provide that evidence in the form of a declaration. Id. at 613.
The Third Hearing on the Motion for a New Trial
¶23 On October 23,2012, two days before the date set for the next evidentiary hearing, defense counsel filed an affidavit stating that he had consulted with a DNA testing expert, Dr. Gregory Hampikian, and concluded that a new trial was warranted. Attached to the affidavit was a brief report by Dr. Hampikian. The report concluded that Olson could, in theory, have contaminated the DNA evidence used to convict Davila, provided that Davila’s DNA was in the Crime Lab (and was handled by Olson) before Olson tested Swab D. It stated, very generally:
In the present case (involving Julio Joseph Davila), Olsen [sic] performed critical DNA tests on evidence, and had access to the key DNA samples used to implicate Mr. Davila. With her well-documented propensity for errors, her work in this case is suspect. While I cannot determine if Mr. Davila’s DNA was in the laboratory at the same time (or before) the evidence samples in this case, it is clear that two evidence samples in this case (the sample taken from the car, and that from the bat) were handled and processed by Ms. Olsen [sic]. If the car sample had Mr. Davila’s DNA, it is possible that Ms. Olsen [sic] mislabeled or contaminated the'samples, so that her finding of Mr. Davila’s DNA on the bat is incorrect. This is concern [ ] based on her well-documented, long-term deficiencies, and the specific mislabeling of samples described in her performance records.
[66]*66If Mr. Davila’s DNA had been in the lab on evidence, or as a reference sample in another case, then the possible routes of contamination are greatly multiplied.
CP at 310-11. The defense provided no other factual data.
¶24 In response, the State filed a certificate from Lorraine Heath, the supervising forensic DNA scientist at the Crime Lab. Most significantly, the certificate stated that (1) during the time that Olson conducted testing for the Jeramie Davis case, “there was never any reference swabs or other items that contained the DNA of Julio Davila in the laboratory” and (2) Olson did not have possession of the steering wheel swabs at the time she tested Swab D, so “[t]here is no possibility that the steering wheel swabs could have contaminated, or been switched with, the baseball bat swab.” CP at 313.
¶25 On October 25, 2012, the trial court held its final hearing on the motion for a new trial. Defense counsel submitted no other evidence and called no witnesses. He did dispute Heath’s assertions that the steering wheel swabs could not have contaminated the bat swab:
Based on [Dr. Hampikian’s] review of the records ... I believe approximately August 16th 2007, is when the bat was tested. Based on Dr. Ha[m]pi[ ]kian’s review of the records, on August 14th, two days prior . . . the automobile, truck, steering wheel swabs were in the lab.
[67]*67So two days prior Ms. Olson had, in the lab, according to Dr. Ha[m]pi[ ]kian’s review of Washington State Patrol records, evidence from the truck, evidence from inside the store.
4 VRP at 618 (Oct. 25, 2012).
¶26 The court apparently accepted the factual assertion that the steering wheel swab was in the Crime Lab before the bat sample was tested, but rejected defense counsel’s argument because the simultaneous presence of several evidentiary items in the Crime Lab was not enough to suggest that cross contamination had occurred:
I think the question that I have, counsel, the fact that there are two or three or four items in the lab, in and of itself, does not say whether or not those items were in proximity at a time and a place where contamination could have occurred. The fact that they are simply in the lab, normally the case that I referenced to you before, they were being tested all at the same time. So it wasn’t just that they were in the lab; they were in close proximity being tested. And as you can see on these reports, there are a number of items on each report that were being tested, four or five items on each report on different dates. But these two items were not tested together. In other words, the bat was tested with a few other items, the steering wheel was tested with a few items, but they were not tested together, and apparently not on the same day. That seems to be the issue to me. ... So the fact that they were both in the lab at the same time, in and of itself, does not mean a whole lot to me. I do not have any evidence that they were necessarily either mingled in the storage area, and it seems pretty clear from Ms. Heath’s certificates that they were not mingled in the testing.
Id. at 619-20. The trial court concluded that these facts were not enough to show contamination in this particular case:
We do know that Mr. Davila’s DNA was not in the lab directly from any prior conviction or prior matter, so the test sample for him came in later. Ms. Heath has provided not only her certificate of where she tested the materials, but also the reports that these materials were not tested together. They were tested on separate days.
[68]*68The first thing that was tested was the bat, and it is the bat that has the DNA. The steering wheel ended up being inconclusive as to whether or not Mr. Davila’s DNA was even on it. So that was inconclusive; it has never been identified specifically as an item that had his DNA. The only thing that has been identified [as] an item that had his DNA was the bat. That was tested first.
While I can appreciate counsel’s concern in this matter about not having known about it, the reality is that the information provided to this jury came from Ms. Heath. She did the testing on it and essentially supported Ms. Olson’s testing. I appreciate the argument that once it is contaminated, it is contaminated and the testing is going to come up the same. Which is one of the reasons that I allowed counsel to go forward on this matter, but I have not seen anything that is anything other than speculation. The facts, as best I can see them, are that these items were tested on different dates. The bat was tested first and the steering wheel was inconclusive and the bat was not inconclusive.
Id. at 623-24.
¶27 Davila appealed, and the Court of Appeals affirmed. State v. Davila, 183 Wn. App. 154, 333 P.3d 459 (2014). In Brady, the United States Supreme Court held that “the suppression by the prosecution of evidence favorable to an accused upon request violates due process where the evidence is material either to guilt or to punishment, irrespective of the good faith or bad faith of the prosecution.” 373 U.S. at 87 (emphasis added). The Court of Appeals held that Olson’s incompetence was favorable to the accused and suppressed by the State—thus meeting the first two Brady requirements. 183 Wn. App. at 167-70. But the court also concluded that the new evidence was not material. Id. at 167-73.
¶28 The Court of Appeals acknowledged that Davila’s defense would have proceeded very differently if defense counsel had known that Olson was fired for incompetence. Specifically, the court noted that the defense could have called Dr. Hampikian to testify about his concerns that [69]*69DNA from the steering wheel swab might have contaminated the bat swab. Id. at 171-72. But it held that Olson’s incompetence was not “material,” under Brady, because “close review of the record establishes little likelihood that her handling of the evidence could have contaminated the evidence at issue.” Id. at 172-73. Just as the trial court had done, the Court of Appeals relied on Olson’s own records to conclude that the steering wheel and bat swabs were tested on different days, “astronomically reducing the possibility of cross contamination.” Id. at 172. Ultimately, the court concluded that evidence of Olson’s incompetence could have been used “for impeachment purposes” but was not material because there was no evidence that Olson had mishandled the DNA samples used to convict Davila. Id. at 173.
¶29 Davila petitioned for review, arguing that the Court of Appeals applied the wrong legal standard when it determined that he had not satisfied Brady’s materiality requirement. State v. Davila, 182 Wn.2d 1002, 342 P.3d 327 (2015).
ANALYSIS
¶30 The Court of Appeals was correct. In order to establish a Brady violation, a defendant must establish three things: (1) “[t]he evidence at issue must be favorable to the accused, either because it is exculpatory, or because it is impeaching,” (2) “th [e] evidence must have been suppressed by the State, either willfully or inadvertently,” and (3) the evidence must be material. Strickler v. Greene, 527 U.S. 263, 281-82, 119 S. Ct. 1936, 144 L. Ed. 2d 286 (1999). The evidence that Olson was inept was both favorable to the defense and suppressed by the State (which had a detailed report on Olson’s professional failings). But, despite the fact that the trial court gave the defense three chances to present evidence in support of materiality, the defense did not do so. On this record, the defense failed to prove materiality.
[70]*70I. The Court of Appeals correctly held that evidence of Olson’s incompetence was favorable to the accused
¶31 As noted above, “ ‘favorable’ ” evidence under Brady includes impeachment evidence as well as exculpatory evidence. Strickler, 527 U.S. at 280 (quoting Brady, 373 U.S. at 87). The Court of Appeals concluded that Olson’s incompetence was favorable to the accused because it “would have opened an area of impeachment that Mr. Davila was unaware of at the time of trial.” Davila, 183 Wn. App. at 168. Citing this court’s decision in Stenson, 174 Wn.2d at 489, it noted the importance of thoroughly cross-examining forensic analysts. We agree with the Court of Appeals on this point.
¶32 The State asserts that Olson’s deficiencies are not truly “exculpatory” since the allegation that “Olson could have contaminated or mislabeled the samples was mere speculation.” Suppl. Br. of Resp’t at 23 (formatting omitted). In support of that claim, it cites United States v. Michaels, 796 F.2d 1112, 1116 (9th Cir. 1986), United States v. Georgiou, 111 F.3d 125, 141 (3d Cir. 2015), cert. filed, No. 14-1535 (U.S. June 25, 2015), and United States v. Andrus, 775 F.2d 825, 843 (7th Cir. 1985). Suppl. Br. of Resp’t at 23-24.
¶33 This authority is distinguishable. In Michaels, the defense sought access to United States “postal inspectors’ rough notes from witness interviews,” speculating that those notes might have proved that the defendant never actually mailed a package of explosives. 796 F.2d at 1115. The court held that the notes were not material because the defendant provided no basis for believing that they actually contained exculpatory information. Id. at 1116.
¶34 Similarly, in Georgiou, the defendant sought notes from two federal Securities Exchange Commission interviews of a witness for the prosecution. 777 F.3d at 141. The court held that there was no basis in the record to suggest that the notes contained any Brady material. Id. And in Andrus, the [71]*71defendant sought to compel discovery of law enforcement witness personnel files. 775 F.2d at 843. The court held that the defendant had no basis' for concluding that the files contained Brady material and was not entitled to discovery based on speculation alone. Id.
¶35 Unlike the situations in Michaels, Georgiou, and Andrus, here we know exactly what the alleged Brady material contains: evidence that Olson made numerous mistakes in her forensic analytical work and that the Crime Lab nevertheless continued to employ her for several years after the mistakes first surfaced.
II. The Court of Appeals correctly held that the State suppressed evidence of Olson’s incompetence
¶36 Under Brady, the prosecution has a duty to seek out exculpatory and impeaching evidence held by other government actors. Kyles, 514 U.S. at 438. Thus, the prosecution “suppresses” evidence, for purposes of Brady, even if that evidence is held by .others acting on the government’s behalf, e.g., police investigators. Id.; see also Strickler, 527 U.S. at 283 n.23 (if prosecution asserts that it complies with Brady through an open file policy, defense counsel may rely on file to contain all Brady materials). As the State now appears to acknowledge, the Crime Lab is an arm of the State whose knowledge is imputed to the prosecution for purposes of Brady. See Giglio v. United States, 405 U.S. 150, 154, 92 S. Ct. 763, 31 L. Ed. 2d 104 (1972) (one federal prosecutor’s knowledge imputed to another federal prosecutor for purposes of Brady suppression inquiry). For this reason, the trial court was wrong to suggest that the prosecution had no obligation to discover or disclose the records pertaining to Olson’s incompetence.3
[72]*72¶37 Nevertheless, the State argues that it could not have “suppressed” evidence of Olson’s incompetence since Olson was listed as a witness for the defense. It contends that the defense listed Olson as a potential expert witness and should therefore have discovered the facts of her termination through the normal expert-witness-vetting process. Suppl. Br. of Resp’t at 22 (“the defense attorneys in the present case had an elementary and fundamental obligation to at least question their listed expert witness, Ms. Olson, before trial about her qualifications and work history with the Washington State Patrol”).
¶38 Davila counters that the defense never listed Olson as an expert witness but instead considered calling her only as a fact witness “regarding ‘chain of custody issues’ and DNA tested from Allen’s truck.” Pet’r’s Second Suppl. Br. Addressing New Issues Raised in State’s Br. at 7 (citing 4 VRP at 575-78 (Aug. 1, 2012); CP at 283).
¶39 The record supports his contention. The defense mentioned Olson only once in the trial court, when it discussed its anticipated cross-examination of Heath:
THE COURT: Mr. Krzyminski, do you have witnesses available for tomorrow if we need them?
MR. KRZYMINSKI: The witness that I would anticipate is Denise Olson, ydur Honor. In my conversations with Mr. Nagy, I believe the testimony that I can elicit from Denise Olson I can get from Lorraine Heath. If that worked out all fine, I would not be calling Denise Olson. But again that’s subject to Ms. Heath. I’m anticipating that she can answer the questions .1 would have because it has to do with the previous tests [performed for Davis’ case]. And [Heath] didn’t work there in 2007, 2008 when these items were being tested hut she does have the reports available.
2 VRP at 398-99 (July 11,2012) (emphasis added). It is clear from this exchange that defense counsel anticipated calling [73]*73Olson only as a fact witness regarding the handling of relevant DNA evidence in Davis’ case. The defense did not, as the State argues, list Olson as an expert witness. Even if it had, the defense would not thereby waive the defendant’s constitutional Brady protections.
¶40 Further, the defense had no reason to investigate Olson’s reasons for leaving the Crime Lab just because she was a possible witness. For this reason, all of the authority that the State cites on the defense’s duty to exercise due diligence in discovering exculpatory evidence is distinguishable.4 The Court of Appeals correctly held that the prosecution suppressed evidence of Olson’s incompetence in the Crime Lab, and of testifying witness Heath’s knowledge of this incompetence, for purposes of the Brady analysis.
III. The Court of Appeals correctly held that evidence of Olson’s incompetence was not material under Brady given the specific facts in this record
¶41 Evidence is material under Brady “ ‘if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different.’ ” Kyles, 514 U.S. at 433-34 (quoting Bagley, 473 U.S. at 682). To satisfy this standard, a defendant need not demonstrate by a preponderance that he would have been acquitted had the suppressed evidence been disclosed. Id. at 434. Instead, he or she must show only that “the government’s evidentiary suppression ‘undermines confidence in the outcome of the trial.’ ” Id. (quoting Bagley, 473 U.S. at 678). There is no separate, additional prejudice inquiry.
[74]*74¶42 The parties’ substantive arguments on materiality-are intertwined with an argument over the applicable standard of review. The defense argues that, at least in this case, we should review both the trial court’s conclusions of law and its findings of fact de novo. The State acknowledges that this court reviews Brady claims de novo, but it contends that deference is owed the factual findings underlying a trial court’s ruling on a Brady claim. Suppl. Br. of Resp’t at 12.
¶43 For the following reasons, we conclude that (a) Brady materiality is a legal question that is reviewed de novo, but the trial court’s underlying factual findings are reviewed for substantial evidence; (b) the trial court applied the correct legal standard when it denied Davila’s Brady claim, even though it used some imprecise language; and (c) on the basis of the available evidence, the trial court did not err in concluding that Item 24 could not have contaminated Swab D.
A. Whether suppressed evidence is “material” is a legal question that is reviewed de novo, but the trial court’s underlying factual findings are reviewed for substantial evidence in the record
|44 In State v. Mullen, 171 Wn.2d 881, 893-94, 259 P.3d 158 (2011), this court stated that it reviews Brady claims de novo. But it did not specifically address the issue of deference to the trial court’s factual determinations—it simply agreed with the trial court’s decision to deny the defendant’s motion for a new trial. Id. at 887, 898-905. In Stenson, 174 Wn.2d at 488—a personal restraint petition— this court explained that Brady materiality is a mixed question of fact and law, which “[w]e review . . . de novo by applying the reference hearing facts to the law and drawing our own legal conclusions.”
¶45 This precedent applies a typical mixed standard of review for Brady claims: the trial court’s legal conclusions about materiality are reviewed de novo, but its underlying [75]*75factual findings are reviewed for substantial evidence in the record.5 This is consistent with federal appellate court precedent on Brady materiality.6 Thus, while the trial court’s factual conclusions receive some deference, the ultimate constitutional question—whether the suppression of evidence deprived the defendant of due process—is reviewed de novo.7
[76]*76
B. The trial court applied the correct legal standard for materiality in this case
¶46 As noted above, Davila argues that the trial court applied the wrong standard for materiality when it denied his motion for a new trial. Specifically, he contends that the trial court should have asked whether Olson’s termination for incompetence would have allowed him to impeach the overall carefulness of the Crime Lab—not whether there was any reasonable possibility that Olson contaminated the evidence used in his case.
¶47 Davila argues that the facts underlying Olson’s termination “undercut the credibility of the person who first handled, extracted, amplified, and tested the DNA samples from the crime scene”—Olson—and “impeached the carefulness of the Crime Lab.” Suppl. Br. of Pet’r at 7. He maintains that the defense would have employed a fundamentally different strategy had it known of these facts. Davila points out that Heath testified to the rigorous accreditation standards that the Crime Lab must meet, including regular audits of every lab scientist’s proficiency. He argues that had the defense known how long the Crime Lab employed a scientist with serious deficiencies like Olson’s, it could have tried to impeach Heath’s testimony by calling an expert to explain the risk of error that Olson posed and investigating the possibility that Heath wanted to downplay that risk.
¶48 But in all of the cases on which Davila relies, the defense actually developed factual data that would support a legal finding of materiality.
¶49 In Benn v. Lambert, 283 F.3d 1040, 1055 (9th Cir. 2002), for example, the prosecution failed to disclose the fact that an informant-witness had a history of lying and was viewed by law enforcement as untrustworthy. And in Amado v. Gonzalez, 758 F.3d 1119, 1139 (9th Cir. 2014), the prosecution failed to disclose the facts that a gang rivalry and the desire to seek favor with probation officers might [77]*77have motivated its primary witness to testify against the defendant. In both of these cases, it was readily apparent what strategy the defense would have pursued had it known of the suppressed impeachment evidence. And, more to the point, it was readily apparent that this strategy would have seriously undermined the prosecution’s theory.
¶50 The other two cases Davila cites are similar, although they deal specifically with forensic evidence. In Aguilar v. Woodford, 725 F.3d 970, 971 (9th Cir. 2013), the prosecution introduced evidence that a police dog had alerted to the scent of the defendant at the scene of the crime. The other evidence against the defendant was weak, and substantial evidence suggested that another person had committed the crime, a shooting. Id. The Ninth Circuit held that the dog’s history of mistaken scent identifications was material under Brady. Id. at 983-85.
¶51 In Stenson, the defendant was convicted of shooting two victims to death. 174 Wn.2d at 487. The State’s case rested in part on forensic evidence showing (1) gunshot residue inside the front right pocket of the defendant’s jeans and (2) blood spatter on the front of the same jeans. Id. at 478. Long after trial, postconviction counsel discovered two pieces of impeaching evidence: (1) a photograph of the lead detective, who testified at trial and supervised the entire investigation, showing him wearing the defendant’s jeans and turning out the right pocket with an ungloved hand and (2) a Federal Bureau of Investigation file revealing that a trainee, and not the expert witness who testified for the State, had actually performed the gunshot residue testing. Id. at 479, 491. After a lengthy evidentiary hearing addressing the handling of the jeans, this court held that the photograph tended to “demonstrate . . . that a key exhibit in the case . . . had [in fact] been seriously mishandled and compromised by law enforcement investigators.” Id. at 492.
¶52 Thus, in both Aguilar and Stenson, the Brady evidence at issue tended to completely neutralize the prosecu[78]*78tion’s most significant evidence. In Aguilar, the Brady evidence allowed the defense to argue that the scent identification was incorrect, like many of the police dog’s other scent identifications had been. And in Stenson, the evidence undermined key scientific evidence in the State’s case.
¶53 We agree with Davila that he could have used the facts of Olson’s ineptitude to generally undermine the Crime Lab’s reputation with the jury,8 but we conclude that the defense failed to meaningfully connect Olson’s ineptitude with the evidence used to convict Davila. Put another way, the defense failed to develop facts showing that Olson’s ineptitude and termination were material in this case. To be sure, the prosecution’s Brady duties “encompass [ ] impeachment evidence as well as exculpatory evidence.” Strickler, 527 U.S. at 280 (emphasis added) (citing Bagley, 473 U.S. at 676). But the effect of any omission must be evaluated cumulatively, Kyles, 514 U.S. at 440, and in the context of the whole trial record, Augurs, 427 U.S. at 112-13. Indeed, if impeachment value alone were sufficient to warrant a new [79]*79trial, there would be no need for the third Brady prong (the materiality inquiry).9
¶54 But in this case, even if Olson’s test results were untrustworthy, the defense still faced the problem of Swab D (the baseball bat swab). Testing by both Olson and Heath revealed the presence of Davila’s DNA on that swab.
¶55 Davila seeks to attribute both of those inculpatory test results to Olson’s incompetence, but he suggests only one theory that could support that attribution: the theory that Item 24 did have Davila’s DNA on it (even though testing of Item 24 by both Olson and Heath was inconclusive for Davila’s DNA), that Olson handled Item 24 before handling Swab D (or handled them simultaneously), and that Olson thereby contaminated Swab D with DNA from Item 24.10 Timing was thus crucial to Davila’s argument on materiality. Yet he never offered any fact or called any witness who could have supported his theory that Olson in fact handled Item 24 before she handled Swab D or that those two items were otherwise mingled in the Crime Lab, [80]*80even though the trial court gave him two continuances to do so. Indeed, the only evidence presented at the Brady hearing supported the opposite conclusion.
¶56 Given the specific facts of this case, the trial court was correct to require some evidence that Item 24 could have contaminated Swab D in order to find materiality. Without that evidence, Olson’s termination for incompetence does not seriously undermine Heath’s testimony that Davila’s DNA was on the handle of the baseball bat used to kill Allen.
¶57 The trial court did err when it stated that Davila needed to prove “on a more likely than not basis, [that] Mr. Davila’s DNA was on something that was in the lab at the relevant time.” 4 VRP at 612-13 (emphasis added) (Oct. 19, 2012); see supra p. 73. But since Davila did not establish any likelihood of contamination at all, under any standard, this misstatement is irrelevant.
C. On the basis of the available evidence, the trial court did not err when it concluded that there was no reasonable probability of contamination
¶58 As noted above, the trial court found that there was no evidence of contamination because Item 24 and Swab D were tested on different days and Swab D was tested first. 4 VRP at 623-24 (Oct. 25, 2012). Davila challenges this finding in two ways.
¶59 First, Davila argues that it does not matter whether Item 24 and Swab D were tested on different days. He contends that Item 24 could have contaminated Swab D solely by virtue of the fact that the two samples were held together in the lab and handled by Olson. Pet’r’s Second Suppl. Br. Addressing New Issues Raised in State’s Br. at 6. Second, Davila argues that the Court of Appeals (and, by extension, the trial court) should not have relied on Olson’s reports to conclude that Item 24 and Swab D were tested on different days since sloppy record keeping was one of the [81]*81reasons that Olson was fired. Pet’r’s Suppl. Br. at 18. We reject both of these arguments.
¶60 To support his argument on cross contamination, Davila offered only Dr. Hampikian’s declaration, which stated that cross contamination could have occurred because Olson, an incompetent scientist, handled both samples. But, as the trial court pointed out at the second hearing on the motion for a new trial, Item 24 could not have contaminated Swab D unless those items were actually commingled—for example, by Olson’s handling them simultaneously or handling Item 24 before handling Swab D—and on this factual question, Dr. Hampikian had no knowledge whatsoever. Thus, his observation that Olson’s incompetence might have led to cross contamination was speculative for purposes of the materiality inquiry.
¶61 In fact, the only evidence regarding the actual handling of Item 24 and Swab D—Heath’s certificate and Olson’s records—supported the State’s argument on materiality: that Olson tested (and therefore handled) Swab D weeks before she tested Item. 24. We agree with Davila that this evidence was not perfect—Heath is the very witness that Davila sought to impeach,11 and Olson’s records are inherently suspect given her propensity for error. But this imperfect evidence was unrebutted. The defense offered no evidence of its own to help the trial court decide if Hampikian’s cross contamination theory had any possible application to the facts of this case. Davila didn’t show where Crime Lab evidence is stored, how it is generally handled, whether that was different in 2007, or what could have occurred in this case to mingle Swab D and Item 24.
[82]*82¶62 The defense has the burden to produce facts in support of its theory, and the defense failed to develop any such facts in this case. Thus, the trial court did not err when it determined that there was no reasonable probability of contamination.
CONCLUSION
¶63 The defense failed to present any evidence in support of its claim that contamination was a real possibility in this case. Based on the defense’s presentation in the Brady hearings, the trial court reached the right conclusion on materiality. We agree with the Court of Appeals’ analysis on all three Brady requirements, and we therefore affirm.
Madsen, C.J., and Johnson, Owens, Fairhurst, Stephens, Wiggins, González, and Yu, JJ., concur.
2 These assertions are not supported by anything in the record before this court. According to that record, Olson reported testing the bat on November 5,2007, and the steering wheel swab on December 4, 2007. CP at 275, 278; see also 3 VRP at 455 (July 12, 2012) (Heath testifying that Olson tested Swab D on November 5, 2007). It is not clear why defense counsel referred to the August dates at the Brady hearing. In his second supplemental brief in this court, Davila sirgues that the August dates are correct and that the “actual testing [of the bat and the steering wheel swabs] was [therefore] performed within two days in August 2007 and the items were held together in the lab.” Pet’r’s Second Suppl. Br. Addressing' New Issues Raised in State’s Br. at 6. But in support of this assertion, Davila cites only to the testimony by defense counsel, excerpted above.