OPINION

Kenneth Olsen was convicted in 2003 of knowingly possessing a biological agent, toxin, or delivery system for use as a weapon, in violation of 18 U.S.C. § 175 . This court affirmed his conviction on direct appeal. See United States v. Olsen, 120 Fed.Appx. 18 (9th Cir.2004). In 2009, Olsen filed a motion to vacate, set aside, or correct his sentence under 28 U.S.C. § 2255 , raising a number of grounds for relief. After holding an evidentiary hearing on one of those grounds, the district court denied the motion. This court granted a certificate of appealability as to four of Olsen’s claims: violation of Brady v. Maryland, ineffective assistance of counsel, juror bias, and cumulative error. We have jurisdiction under 28 U.S.C. §§ 1291 and 2255. We affirm.

I. Background

Kenneth Olsen was an employee of Agi-lent Technologies, Inc., in Liberty Lake, Washington, where he worked as a computer support technician. In August of 2001, a co-worker discovered that a document labeled the Terrorist Encyclopedia, which referenced explosives and other disturbing matters, had been printed on a shared office printer. Agilent security identified Olsen as the person who printed the document. Further inquiry revealed that from his office computer Olsen had extensively browsed internet websites involving poisons and killing, specifically conducting internet searches for phrases like “undetectable death pill.”

Agilent security personnel and counsel interviewed Olsen, who offered innocuous explanations for his internet browsing that were not deemed credible. He claimed, for instance, that he searched for information on poison to help his son’s Boy Scout troop avoid poisonous berries, and that he viewed anarchist websites involving toxic chemicals because he wanted to learn to make safe household cleaners. Olsen was immediately placed on leave and escorted off the property without being allowed to return to his cubicle.

The next day, the company decided to fire Olsen, and it sent employees to clear out the personal items from his cubicle. Within Olsen’s filing cabinets, the employees discovered a wealth of internet printouts and books on poisons and methods of harming people and exacting revenge. They also found test tubes and a wide assortment of other chemistry paraphernalia. Alarmed by their discovery, the employees placed the items into several boxes and took them to the Agilent security team, which in turn contacted the Spokane County Sheriffs Office. A deputy sheriff responded and took custody of the items, and a detective later took them to the Washington State Patrol (“WSP”) crime laboratory. There, the items were examined by WSP forensic scientist Arnold Melnikoff.

Melnikoff tested the items for the presence of various substances. Among the many items he examined were two test tubes containing an oily residue, a metal cup with a small amount of white “cakey” residue stuck inside of it, a glass bottle with similar residue caked on the bottom of the bottle, a bag of beans later identified as castor beans, and several bottles of medicine, including a bottle of Equate-brand allergy capsules. Melnikoffs discovery that the test tubes contained castor oil, and that the beans appeared to be castor beans, alerted him that some of the substances might contain ricin, a highly deadly poison derived from castor beans. Because the WSP lab did not have the ability to test for ricin, Melnikoff contacted the FBI and arranged for it to analyze the test tubes, metal can, glass jar, and beans. Melnikoff then individually sealed each of the items for the FBI.

The FBI in turn sent the test tubes, metal can, and glass jar to the United States Army Medical Research Institute of Infectious Diseases (“USAMRIID”), which subsequently notified the FBI that each tested positive for ricin. The FBI then took possession of the remaining items from the WSP lab and sent them to US-AMRIID. Twelve items from Olsen’s cubicle ultimately tested positive for ricin. With respect to the Equate capsules, US-AMRIID tested twelve pills for ricin, finding that one had a “high concentration” of the poison while three others also tested positive but “below the level of quantization.” Because the capsules had to be liquified for testing, however, it could not be determined whether the ricin was inside the tainted capsules or on their surfaces.

Independent of Melnikoffs contact with FBI laboratory personnel, the Spokane detective handling the case had already consulted the FBI about the matter. When USAMRIID confirmed that the first batch of items tested positive for ricin, the FBI took over the investigation.

Olsen was indicted in July 2002 for knowingly possessing a biological agent, toxin, or delivery system for use as a weapon, in violation of 18 U.S.C. § 175 . In April 2003 a second indictment was returned, charging Olsen with possessing a chemical weapon in violation of 18 U.S.C. § 229 . After a twelve-day jury trial in July 2003, Olsen was found guilty of both charges.

At trial, Olsen did not contest the fact that he produced and possessed ricin, but only that he did so “for use as a weapon,” 18 U.S.C. § 175 , and that he intended to possess the ricin as a “chemical weapon,” 18 U.S.C. § 229 . Defense counsel attributed Olsen’s actions to “an irresponsible sense of curiosity” about “strange and morbid things.” The prosecution, on the other hand, presented evidence that Olsen methodically researched numerous undetectable means of killing someone, investigated not only how to produce poisons but also various means of delivering them to a victim, purified the ricin he produced to enhance its toxicity, performed mathematical calculations to determine the maximum doses of certain common medications for a 150-pound person, and spiked the aforementioned Equate capsule with the ricin he produced.

Arnold Melnikoff did not provide expert testimony at trial, but he testified about his handling of the items recovered from Olsen’s cubicle, the tests he performed on them, the reasons he contacted the FBI for assistance, and his packaging of the items for transfer to the FBI.

This court affirmed Olsen’s conviction on direct appeal, concluding, among other things, that the government provided “copious evidence” of Olsen’s intent to use the ricin as a weapon. The court remanded for resentencing, however, due to an error in the application of the Sentencing Guidelines. Olsen, 120 Fed.Appx. at 19 . Olsen eventually was sentenced to 121 months of imprisonment and 60 months of supervised release, and was ordered to pay restitution. He appealed the revised sentence, which this court affirmed. See United States v. Olsen, 280 Fed.Appx. 624 (9th Cir.2008).

In 2009, Olsen filed his Section 2255 motion, raising numerous grounds for relief. The district court held an evidentiary hearing on one of these issues, Olsen’s claim of juror misconduct. Although the court initially decided to hold an evidentia-ry hearing on a second issue—a Brady v. Maryland claim premised on the suppression of information about Arnold Melni-koff—the court later determined that no hearing was necessary. The court did allow the parties to file briefs, however, along with exhibits that had been prepared for the evidentiary hearing, including several depositions and affidavits. The district court denied Olsen’s Section 2255 motion in part on July 15, 2010, rejecting his juror bias claim. On November 9, 2010, the court rejected the remainder of Olsen’s claims and denied the motion in its entirety. The district court declined to issue a certificate of appealability on any issue, but this court granted a certificate with respect to the four issues presented here.

II. Legal Standards

We review de novo the district court’s denial of a 28 U.S.C. § 2255 motion. United States v. Aguirre-Ganceda, 592 F.3d 1043, 1045 (9th Cir.2010). The denial of an evidentiary hearing is reviewed for abuse of discretion. Mendoza v. Carey, 449 F.3d 1065, 1068 (9th Cir.2006). A district court’s Brady determinations and all other questions of law are reviewed de novo. United States v. Kohring, 637 F.3d 895, 901 (9th Cir.2011).

III. Brady Claim

Olsen contends that the prosecution suppressed information seriously undermining the honesty and professional competence of Arnold Melnikoff, the WSP forensic scientist who first examined the items recovered from his cubicle that later were found to be contaminated with ricin. Had this information been available to Olsen at trial, he maintains, it could have been used to rebut the government’s argument that Olsen spiked the Equate pill with ricin: in light of Melnikoffs record of laboratory incompetence, Olsen would have had a basis for arguing that, instead, Melnikoff inadvertently contaminated the pill and did not testify truthfully about the laboratory procedures he followed when examining Olsen’s items. In Olsen’s view, the contaminated Equate capsule was critical to the government’s case because it offered powerful evidence that Olsen possessed the ricin to harm someone, as required for conviction under 18 U.S.C. § 175 , rather than merely out of a morbid fascination with poisons.

Although Melnikoff did not perform the tests that identified Olsen’s possessions as containing ricin (nor testify as an expert at trial), he handled and extensively manipulated the items before they were transferred to FBI custody. An expert affidavit supporting Olsen’s Section 2255 motion concludes, based on the evidence presented at trial, that it is “highly probable” that Melnikoff contaminated the Equate capsule with ricin after handling the other, ricin-positive items. Melnikoff testified that he examined the Equate capsules not by individually removing them from the bottle with forceps, but rather by dumping them onto his laboratory bench, albeit on “a sheet of clean lab paper,” after he had examined other items on the same bench— which included scraping ricin-positive powder from some of these items. Melnikoff also does not appear to have changed gloves between handling each item that he examined, although his testimony was equivocal on this point.

A. Background

Before Melnikoff joined the WSP in 1989, he was the head of the Montana State Crime Laboratory, where his work included conducting hair sample analyses. A defendant named Jimmy Ray Bromgard, who was convicted in Montana of rape based largely on Melnikoffs hair sample testimony, was later exonerated by DNA testing and released in 2002. An Innocence Project advocate who had worked on the Bromgard matter wrote to the Washington State Attorney General about Mel-nikoffs expert testimony in that case and urged a review of his subsequent work in Washington.

The WSP responded by launching an internal investigation of Melnikoff in October 2002 that included “a review of current work performance.” A letter advised Mel-nikoff of the allegations against him: that while employed in Washington, he “may have engaged in misconduct involving courtroom testimony and/or case analysis” and that he “may have misrepresented himself during the original employment process when he applied for a position with the Washington State Patrol.” Melnikoff was placed on administrative reassignment during the investigation. In January 2003 two additional charges were added, stemming from Melnikoffs expert testimony in a different Montana case where the defendant later was exonerated. These charges accused Melnikoff of offering statistical conclusions regarding hair sample identifications that were not consistent with scientific principles and of substantially overstating the number of cases in which he had conducted hair analyses.

Documents generated during a WSP internal investigation are not available to the public or even to the employee being investigated until the WSP’s administrative decisionmaker issues his or her preliminary findings, in the form of an “Administrative Insight.” The Administrative Insight in the Melnikoff investigation was issued on August 4, 2003, after the conclusion of Olsen’s trial. By May 2003, however—nearly two months before the trial began—the investigative file for Melnikoff had already been completed and provided to the decisionmaker for review. This investigative file contained more than unsubstantiated allegations concerning the quality of Melnikoffs laboratory work and his reliability as a witness. Instead, the investigative file contained comprehensive factual findings that investigators reached after scrutinizing Melnikoffs laboratory work and his testimony as an expert witness in both Montana and Washington. Among other items, this file contained three separate evaluations of Melnikoffs work in Montana and Washington, solicited for the investigation and conducted by peers and other experts in forensic laboratory analysis. These evaluations comprised an assessment of Melnikoffs work in the Bromgard case by a panel of experts, an assessment of Melnikoffs work in another Montana case by a forensic scientist, and, most notably, an assessment of Melnikoffs work in 100 recent Washington cases, along with his courtroom testimony in twelve Washington cases, by a panel of Washington forensic chemists. This last assessment—the peer review of Melnikoffs work in Washington—is referred to by the parties as the “Johnston Report”.

The conclusions of these expert assessments were highly critical, calling into question Melnikoffs diligence and care in the laboratory, his understanding of the scientific principles about which he testi fied in court, and his credibility on the witness stand. The Johnston Report, which reviewed Melnikoffs Washington casework, concluded that his laboratory methods “seemed to be built around speed and shortcuts.” The panel noted the presence of unexplained contaminants in his laboratory, among other findings.

The lead prosecutor in Olsen’s trial, Assistant U.S. Attorney Earl Hicks, was aware of the investigation into Melnikoff. In April of 2003, Hicks contacted the WSP laboratory director to inquire about the status of the investigation and to request any Brady material that might exist. Hicks was referred to Elizabeth Brown, an Assistant Attorney General for the State of Washington and legal counsel for the WSP. Although Hicks attempted to reach Brown, and the two repeatedly left voice-mail messages for each other, they never actually spoke or communicated in writing about Hicks’s Brady inquiry.

Because Hicks never spoke with AAG Brown, it appears that he never learned that an investigative file had been completed for Melnikoff in May 2003 nor anything about its contents. Olsen’s defense counsel never received the file or any of the documents it contained, including the critical assessments of Melnikoffs laboratory work—counsel had possession only of the original charging documents laying out the allegations against Melnikoff and informing him that an investigation was being conducted.

During a pretrial hearing in late June 2003, AUSA Hicks informed the district court about the internal investigation into Melnikoff and acknowledged that he needed to “follow up and contact the State of Washington again.” But Hicks emphasized that no findings of any wrongdoing had yet been made. Hicks also characterized the WSP investigation—inaccurately, as it turns out—as pertaining solely to Melnikoffs work conducting hair analyses in Montana during the 1980s and early 1990s. Olsen’s counsel explained that she had been “requesting materials about Mr. Melnikoff for quite some number of months now” based on her awareness of the investigation, his placement on administrative reassignment, and the fact that “his cases were being peer reviewed.” Her concern over the internal investigation material, she explained, arose from her knowledge that Melnikoff did some “testing and manipulation” of Olsen’s items and the possibility, in light of the allegations against him, of “cross contamination” if his handling of the evidence was not “appropriate and done with scientific standards.”

The matter was revisited during trial on July 3, 2003, just before Melnikoff testified. By then the prosecutor had provided those documents within his possession to both the court and defense counsel, but these documents did not include Melni-koffs investigative file or any of its contents such as the Johnston Report. The district court ruled that Melnikoff could not be cross-examined about the WSP investigation, based in part on the court’s erroneous understanding that “[t]he only issue here involved at all is whether or not there was some inaccuracy regarding his testimony in Montana about comparing hair samples on rape and homicide cases” and because “there is nothing in here that I see that indicates that there was any problem at all during ... his tenure with the state of Washington.” The court stated that because no findings had yet been made by the WSP decisionmaker, and because the investigation related solely to hair sample analysis in Montana, it would be “unfair to Mr. Melnikoff to allow counsel to delve into this issue, which isn’t at all relevant, and [does not] appear to in any way involve anything, as I understand it, his participation in this case would involve.” “I know of no indication,” the court reiterated, “that he acted inappropri ately while with the State of Washington. There is no indication that he ... didn’t operate properly under lab techniques.”

Melnikoff subsequently testified before the jury about his handling and examination of the items recovered from Olsen’s cubicle. Defense counsel was permitted to elicit from Melnikoff only the fact that during his career complaints had been filed against him. On redirect, the prosecutor established that these complaints had been two or three in number. The prosecutor then asked Melnikoff if he had ever been disciplined or “ever been found to have done anything incorrectly,” to which Melnikoff answered “no.”

On August 4, 2003, after Olsen had been found guilty, the WSP decisionmaker issued his Administrative Insight, finding all the charges against Melnikoff to be proven. After Melnikoff was allowed to respond to the preliminary findings, they were made final, and Melnikoff was fired. These decisions were upheld on various levels of administrative and judicial review.

B. Brady Standards

“The three elements of a claim for a Brady violation are that ‘[t]he evidence at issue must be favorable to the accused, either because it is exculpatory, or because it is impeaching; that evidence must have been suppressed by the State, either willfully or inadvertently; and prejudice must have ensued.’ ” Gentry v. Sinclair, 693 F.3d 867, 887 (9th Cir.2012) (quoting Strickler v. Greene, 527 U.S. 263, 281-82 , 119 S.Ct. 1936 , 144 L.Ed.2d 286 (1999)). We find that the first element, favorability, has been satisfied in this case. We have no need to address the more difficult question presented here by the second element, suppression, because we conclude that the third element, materiality, has not been satisfied and that Olsen therefore is not entitled to relief.

1. Favorability

Evidence is favorable to the accused under Brady if it has either exculpatory or impeachment value. Gonzalez v. Wong, 667 F.3d 965, 981 (9th Cir.2011) (citing United States v. Bagley, 473 U.S. 667, 676 , 105 S.Ct. 3375 , 87 L.Ed.2d 481 (1985)). The prosecution must disclose materials that are potentially exculpatory or impeaching. See Schad v. Ryan, 671 F.3d 708, 715 (9th Cir.2011); Hovey v. Ayers, 458 F.3d 892, 918 (9th Cir.2006).

The contents of the WSP internal investigation file clearly were favorable to Olsen. The file contained three independent assessments completed by different experts and peers calling into question Melnikoffs care in the laboratory and his fidelity on the witness stand. The Johnston Report, which reviews Melnikoffs lab work in 100 recent Washington cases, criticizes his reliance on “speed and shortcuts,” and while it primarily focuses on problematic identifications of chemical substances, it also cites unaddressed contamination of laboratory materials, finding in one instance that “a defendant could have been prejudiced by an inaccurate test,” and in another that Melnikoffs notes were contaminated with “reddish brown powder ... probably from the fingers.” As evidence that Melnikoffs lab work was characterized by sloppiness and haste, the report could have supported a defense theory that Melnikoff inadvertently contaminated the Equate capsule with ricin, undermining one piece of evidence of Olsen’s harmful intent.

The Johnston Report also examined Melnikoffs testimony in twelve Washing ton cases, noting “small misstatements made in a number of testimonies,” “a tendency for conclusions to become stronger as the case developed, from notes to written report to testimony,” and testimony that was either unsupported by the data or outside Melnikoff s field of expertise. Although these findings largely bear on Mel-nikoffs willingness to offer unwarranted scientific conclusions, they also speak to his truthfulness on a more general level, by suggesting a proclivity to shade his testimony in favor of the government’s case. As such, they could have been used to question the accuracy of his account about the care with which he examined Olsen’s items and thus call into question his credibility as a witness. See Gentry, 693 F.3d at 888 (evidence showing that the lead detective “was fired from his previous job for misconduct and that he had lied to obtain search warrants in other cases” was favorable “as it could have been used to impeach the credibility of [the detective] for truthfulness”) (citing Bagley, 473 U.S. at 676 , 105 S.Ct. 3375 ); Carriger v. Stewart, 132 F.3d 463, 479-80 (9th Cir.1997) (en banc) (evidence from corrections file of prosecution witness bearing adversely on his credibility should have been disclosed under Brady).

Echoing the district court, the government maintains that the contents of Melni-koffs investigative file, including the Johnston Report and the other peer evaluations of his work, were not favorable to Olsen because the WSP decisionmaker had yet to make any findings. That proposition, for which the government offers no support, is at odds with the case law in this circuit, which repeatedly has held materials from ongoing investigations to be favorable under Brady. See, e.g., Kohring, 637 F.3d at 903, 905 (e-mails, memoranda, police reports, handwritten notes, and file regarding ongoing investigation into government witness); United States v. Price, 566 F.3d 900, 903 (9th Cir.2009) (information respecting witness’s arrests, not convictions). Indeed, information bearing adversely on the credibility of a prosecution witness is favorable under Brady regardless of whether it was part of any investigation at all. See, e.g., Gonzalez, 667 F.3d at 976, 981 (reports prepared by prison psychologists undermining credibility of government witness); United States v. Alvarez, 358 F.3d 1194, 1206-07 (9th Cir.2004) (exculpatory and impeachment material from probation officer’s files). The fact that the materials at issue here were generated because of an internal investigation is irrelevant. In the government’s view, apparently, no matter what the investigative file contained—evén perhaps a sworn affidavit by Melnikoff himself admitting that he contaminated Olsen’s items with ricin—this evidence would not be favorable under Brady until the administrative decisionmaker concluded that such conduct violated WSP regulations. This position is untenable under Brady, and the government’s tenacious adherence to it is mystifying.

2. Suppression

In order for a Brady violation to have occurred, the favorable evidence at issue must have been suppressed by the prosecution, Price, 566 F.3d at 907 (citing Strickler, 527 U.S. at 281 , 119 S.Ct. 1936 ), and suppression may be either intentional or inadvertent. Strickler, 527 U.S. at 288 , 119 S.Ct. 1936 ; Schad, 671 F.3d at 715 . An “innocent” failure to disclose favorable evidence constitutes suppression even where there is no allegation that the prosecutor acted “willfully, maliciously, or in anything but good faith”—“sins of omission are equally within Brady’s scope.” Price, 566 F.3d at 907 -08 (citing Benn v. Lambert, 283 F.3d 1040, 1053 (9th Cir.2002)); accord United States v. Pelisamen, 641 F.3d 399, 408 (9th Cir.2011) (citing Strickler, 527 U.S. at 281-82 , 119 S.Ct. 1936 ). “Brady has no good faith or inad vertence defense.” Gantt v. Roe, 389 F.3d 908, 912 (9th Cir.2004).

This case presents the complex question of whether suppression occurs under Brady and Kyles v. Whitley, 514 U.S. 419 , 115 S.Ct. 1555 , 131 L.Ed.2d 490 (1995), when a federal prosecutor does not obtain or reveal information favorable to the defendant that is contained in a state internal investigation file. We need not reach that issue here, however, because we conclude that the information contained in Melnikoffs WSP internal investigation file is not material under Brady.

3. Materiality

Even if evidence favorable to the defendant has been suppressed or not disclosed by the prosecution, there is no true Brady violation unless that information is material. Strickler, 527 U.S. at 281-82 , 119 S.Ct. 1936 . The Supreme Court and courts of appeals have found evidence to be “material” when “there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different.” Maxwell v. Roe, 628 F.3d 486, 509 (9th Cir.2010) (quoting Strickler, 527 U.S. at 280 , 119 S.Ct. 1936 ). “A reasonable probability is one that is sufficient to undermine confidence in the outcome of the trial.” Id. (citing Kyles, 514 U.S. at 434 , 115 S.Ct. 1555 ). “The question is not whether the defendant would more likely than not have received a different verdict with the evidence, but whether in its absence he received a fair trial, understood as a trial resulting in a verdict worthy of confidence.” Strickler, 527 U.S. at 289-90 , 119 S.Ct. 1936 (quoting Kyles, 514 U.S. at 434 , 115 S.Ct. 1555 ); see Hovey, 458 F.3d at 916 . Reversal of a conviction or sentence is required only upon a “showing that the favorable evidence could reasonably be taken to put the whole case in such a different light as to undermine confidence in the verdict.” Williams v. Ryan, 623 F.3d 1258, 1274 (9th Cir.2010) (quoting Kyles, 514 U.S. at 435 , 115 S.Ct. 1555 ). This necessarily is a retrospective test, evaluating the strength of the evidence after trial has concluded.

To be considered material under Brady, the undisclosed, favorable evidence must either be admissible exculpatory evi dence or be impeachment evidence, which “need not have been independently admissible to have been material.” Carriger, 132 F.3d at 481 ; see also Price, 566 F.3d at 911-12 . Impeachment evidence is material “because ‘if disclosed and used effectively, it may make the difference between conviction and acquittal.’ ” Carriger, 132 F.3d at 481 (quoting Bagley, 473 U.S. at 676 , 105 S.Ct. 3375 ). “Evidence can be ‘used to impeach’ a witness even if the evidence is not itself admissible, even to impeach”—a written statement, for instance, that contradicts a witness’s testimony but is inadmissible as hearsay could still be used as a prior inconsistent statement to cross-examine the witness. Paradis v. Arave, 240 F.3d 1169, 1179 (9th Cir.2001); Kohring, 637 F.3d at 904 . Inadmissible evidence that could have led to the discovery of admissible evidence also may qualify as material under Brady, although this circuit has not conclusively resolved the issue. Price, 566 F.3d at 911-12 ; Paradis, 240 F.3d at 1178-79 .

As noted, the materials contained in Melnikoff s WSP internal investigation file could have been employed by Olsen’s defense in at least two ways. First, defense counsel could have cross-examined Melni-koff about the reports contained in the file that indicate his lack of total candor on the witness stand and a tendency to shade his testimony in a manner buttressing the strength of the government’s case. See Kohring, 637 F.3d at 905-07 (concluding that police department files documenting investigation into government witness for offenses calling into question his honesty could have been used to impeach witness). This line of inquiry could have been employed to undercut Melnikoff s account of the care with which he handled Olsen’s possessions and hence his credibility. Second, the Johnston Report—which details Melnikoff s slipshod work in the laboratory—could have been used to further advance the theory that it was Melnikoff who contaminated the Equate capsule with ricin. The defense could have pursued this theory by questioning Melnikoff about the report.

The question under Brady is whether there is a reasonable probability that, had the Melnikoff information been disclosed, it would have led to a different result. Maxwell, 628 F.3d at 509 . “The mere possibility that an item of undisclosed information might have helped the defense, or might have affected the outcome of the trial, does not establish ‘materiality’ in the constitutional sense.” Barker v. Fleming, 423 F.3d 1085, 1099 (9th Cir.2005) (quoting United States v. Croft, 124 F.3d 1109, 1124 (9th Cir.1997)). “For purposes of determining prejudice,” therefore, “the withheld evidence must be analyzed ‘in the context of the entire record.’ ” Benn, 283 F.3d at 1053 (quoting Agurs, 427 U.S. at 112, 96 S.Ct. 2392 ).

Upon a careful review of the record, we conclude that there is no reasonable probability that the verdict would have been different if the favorable evidence had been disclosed. Even if Melnikoffls credibility as a witness had been totally destroyed, we are confident beyond doubt that the jury would have found Olsen guilty, based on the overwhelming evidence presented by the government that he intended to use the ricin he possessed as a weapon. Despite the government’s failure to obtain and disclose Melnikoffs internal investigation file, therefore, Olsen received “a trial resulting in a verdict worthy of confidence.” Gentry, 693 F.3d at 888 (quoting Kyles, 514 U.S. at 434 , 115 S.Ct. 1555 ).

Because Olsen did not contend that he did not produce or possess ricin, the trial focused largely on the issue of intent: did he intend to possess the ricin as a chemical weapon or to use the ricin as a weapon? The contaminated Equate pill was one piece of evidence supporting Olsen’s intent, and to be sure, the prosecution highlighted this piece of evidence—mentioning the capsule in its opening statement, devoting attention to it at various points during its case-in-chief, and emphasizing it during closing argument. Far from being the “lynchpin” of the prosecution’s case, however, as Olsen contends, the Equate pill was, as the government avers, “simply one more layer in an already overwhelming case against the Defendant.” Over the course of the twelve-day trial, the prosecution supplied devastating evidence about Olsen’s extensive research into poisons and killing indicating that these efforts were not prompted by mere curiosity but by a methodical effort to find an undetectable means of ending a victim’s life.

The prosecution demonstrated not only that Olsen produced a substantial amount of ricin—enough to kill at least 75 people—but more tellingly that Olsen carried out extra steps to purify the ricin that he produced, enhancing its toxicity and deadliness. While ricin exists at 3 percent to 5 percent purity in its initial state, Olsen refined the ricin in his test tubes to between 18 and 19 percent purity.

The bulk of the government’s case, however, rested on the evidence of intent amply provided by Olsen’s internet research over the course of a year. This evidence simply did not comport with the defense theory at trial of an individual motivated only by curiosity about poisons. Instead, it showed that Olsen carried out a prolonged investigation into a variety of means to inflict harm on someone, with an ever-present focus on avoiding detection, during which he viewed materials that discussed not just how to create or obtain poisons but how to administer them to a victim. Ricin was but one of the deadly substances that Olsen investigated during the course of this research, although it was the one on which he focused the most attention, and ingestion of poison was only one of the methods of delivery discussed in the resources he accumulated. The government’s case did not in any way hinge on the theory that Olsen actually planned to use the spiked Equate capsule, and the capsule was just one indication of intent to which the prosecution pointed.

Government investigators examined over 20,000 pages of internet proxy logs from Olsen’s computer—records of the internet websites that he viewed from the computer. A lengthy summary of those records, exceeding 200 pages, was introduced in evidence at trial and was discussed in detail during the testimony of FBI agents. This summary was combined with an illustrative timeline of Olsen’s internet research, allowing the prosecution to show the evolution of his activity over the course of months.

From this evidence, the jury learned that Olsen had repeatedly viewed and in some instances printed out or purchased works with titles such as “How to Kill,” “Silent Death,” “Getting Even,” and the “Poisoner’s Handbook,” along with countless websites offering detailed instructions on both the production and delivery of poisons. The prosecution could not, of course, directly prove why Olsen consulted these materials or on what specific information in each source he focused. The proxy logs, however, recorded not only the websites that Olsen viewed but the actual words that he typed into internet search engines to locate materials of interest to him. As the prosecution characterized this evidence in closing argument: “What is unique about the evidence in this case is we have captured a thought process. We have captured his own words.” The proxy logs revealed, for instance, that at various times Olsen performed searches for the phrases “silent killers,” “death by poison,” “tasteless poison,” “hidden poison,” “undetectable poisons,” “untraceable poisons,” “painless death,” “untraceable death pill,” “suicide pill overdose,” “how to get ingredients for suicide pill,” “deadly sleeping pills,” and “common ingredients for death by sleep.”

Ricin was but one of the deadly substances that Olsen investigated during the course of his wide-ranging and nefarious explorations. For instance, he also acquired information about nicotine, which in its pure form is highly poisonous. Along with viewing articles on nicotine poisoning, Olsen performed searches for “acquiring liquid nicotine,” “making liquid nicotine,” “distilling nicotine,” “how do you extract nicotine from tobacco,” “nicotine isolation,” “how do you get pure nicotine,” and “pure nicotine buy.” He also viewed websites discussing lethal doses of nicotine and nicotine poisoning symptoms. A pervasive theme in his internet searches, however, was lack of detection, and the jury also learned that ricin is considered an undetectable poison whose deadly symptoms mimic the ailments of pneumonia and flu.

The prosecution also presented evidence from Olsen’s internet browsing history showing that he viewed and searched for websites dealing specifically with the delivery of poison to a victim. Describing this evidence, the district court observed that Olsen “did much research on poisons, how to administer them, in what doses, and to be undetected.” The prosecution did not focus solely on Olsen’s interest in the delivery of ricin by ingestion and certainly not on the spiking of pills. To the contrary, the government repeatedly emphasized that ricin can be delivered to a victim through inhalation, injection by syringe, or ingestion. The prosecution showed that needles and syringes were found in Olsen’s cubicle along with his ricin-related items, and that some of the ricin in his cubicle was in powder form, which an FBI agent testified could be inhaled or “sprinkled anywhere” for purposes of ingestion. Olsen’s interest in such methods of delivery was further demonstrated by the fact that he conducted searches for phrases like “inhaling ricin,” “atomizing a powder,” and “spraying a powder.”

The jury heard an entire line of testimony about Olsen’s extensive research into “knock out drugs,” such as the so-called date-rape drug, and about how ricin could be administered without detection by first rendering the victim unconscious with such a drug and then injecting the victim or causing him or her to inhale the ricin. As the jury learned, Olsen also researched the maximum doses of certain common antihistamine and sleeping-pill drugs—including Equate—which if taken in excessive amounts can render a person unconscious. The government introduced notes in Olsen’s handwriting in which he listed the maximum doses, in milligrams, for four of these medications. Perhaps most incriminating of all, within these same handwritten notes Olsen mathematically calculated the weight in kilograms of a 150-pound person. The jury learned that Olsen’s wife, the woman with whom he was having an affair, and his former supervisor all weighed around 150 pounds.

All of this evidence was prominently highlighted by the prosecution as evidence of intent during trial and closing argument. The record simply does not support Olsen’s claim that the presence of ricin in the Equate capsule was “pivotal” to the prosecution’s case. The contaminated pill was but one piece of evidence put before the jury for which no innocent explanation was plausible. It is likely for this reason that the district court judge who presided over the trial stated in the course of rejecting Olsen’s Brady claim that the Equate capsule “was not that significant to the outcome of the trial,” and that “even if [the] allegation of cross-contamination by Mr. Melnikoff was conclusively proven as true, it would not have affected the outcome of the trial.”

Had the prosecution obtained and disclosed the evidence from Melnikoff s WSP internal investigation file, the most that Olsen’s defense counsel could have accomplished with it would have been to raise a question about whether Olsen or Melnikoff contaminated the Equate pill with ricin and how credible Melnikoff was in describing the care he took in the laboratory. Viewing this evidence “in the context of the entire record,” Berm, 283 F.3d at 1053 (quoting Agurs, 427 U.S. at 112 , 96 S.Ct. 2392 ), we find no reasonable probability that, had the evidence been disclosed, the result of the proceeding would have been different. See Maxwell, 628 F.3d at 509 .

IV. Ineffective Assistance of Counsel

Olsen contends that although his trial attorneys requested Brady material about Melnikoff from the prosecutor, they rendered ineffective assistance under Strickland v. Washington, 466 U.S. 668 , 104 S.Ct. 2052 , 80 L.Ed.2d 674 (1984), by not attempting to obtain such material independently, through Washington’s public disclosure laws, by seeking a subpoena from the district court, or through other means.

In our view, the existing record strongly suggests that none of the alternative means of obtaining Melnikoff s internal investigation file would have been successful, and that his trial counsel did nothing wrong. We need not resolve this matter, however, because “Brady materiality and Strickland prejudice are the same.” Gentry, 693 F.3d at 889 (citing Bagley, 473 U.S. at 682 , 105 S.Ct. 3375 ); see United States v. Spawr Optical Research, Inc., 864 F.2d 1467 , 1472 n. 6 (9th Cir.1988) (“The Strickland standard for prejudice has been considered to impose virtually the same burden on the defense as the standard for materiality in Brady claims.”). If the withheld information does not constitute a Brady violation for lack of materiality, its absence likewise will not support an ineffective assistance of counsel claim. Gentry, 693 F.3d at 889 . Because we have concluded that Olsen’s lack of access to the impeachment evidence about Melnikoff had no material effect on his defense under Brady, “that analysis is dispositive of the prejudice prong of an ineffective assistance claim based on the same evidence.” Id. Olsen’s claim of ineffective assistance therefore cannot succeed.

Y. Juror Bias

Olsen’s next claim is that his trial was tainted by the presence of a biased juror who, unbeknownst to the parties and the district court, had extensive prior knowledge about the case and had repeatedly expressed a belief in Olsen’s guilt. Olsen maintains that this juror secured a spot on the jury by concealing his prior knowledge during voir dire and by lying about his own background in his juror questionnaire. The district court held a two-day eviden-tiary hearing on this claim, after which it concluded that the juror was not biased against Olsen and that his inaccurate juror questionnaire responses were the result of honest oversights. We affirm the district court’s findings and conclusions.

A. Background

One of the jurors in Olsen’s trial was a man named Kenneth Leavitt. During voir dire, Leavitt did not speak up when the prospective jurors were asked whether anyone had “any particular exposure or knowledge” about the case. Defense counsel later asked those prospective jurors who had indicated having knowledge of the case to re-identify themselves and elaborate on their knowledge. Although Leav-itt had not responded earlier, he then identified himself and explained: “Just months and months ago, a little news blurb on one of those news break shows, something about ricin or whatever.” Leavitt did not respond, however, when defense counsel inquired of the individuals who had heard about the case: “Have you talked about this case just in general terms with other people simply because it was an item of news interest? Is there anyone here who has done that?” Nor did he respond when counsel asked of the same group whether any of them had formed any opinions about the case. When Leavitt later was questioned individually by the trial judge, he said he knew of no reason why he could not be impartial.

Leavitt previously was a business associate of an individual named Kevin Ryan, whose wife Jeanette Ryan was a friendly acquaintance of Olsen. Olsen worked as a massage therapist in addition to his computer-related employment, and Jeanette Ryan had gone to him for massage therapy for several years. Ms. Ryan attended portions of Olsen’s trial, which was held near her office, to show support for him and his family. On the first day that she attended the trial, she observed Leavitt—whom she had met a few times because of his work with her husband—in the jury box. That night, she mentioned this observation to her husband. Mr. Ryan, as he later testified at the evidentiary hearing, was “absolutely flabbergasted” and “sick to [his] stomach” by the revelation because, he testified, he and Leavitt had had multiple conversations about the Olsen case when it was in the news, during which Leavitt expressed an unwavering belief in Olsen’s guilt and demonstrated familiarity with the evidence against him.

The Ryans’ concerns about juror Leavitt were brought to the attention of Olsen’s wife and then his counsel, although it appears that the full extent of Mr. Ryan’s allegations regarding his conversations with Leavitt became known to counsel only after sentencing.

B. Discussion

“The Sixth Amendment guarantees criminal defendants a verdict by an impartial jury.” United States v. Martinez-Martinez, 369 F.3d 1076, 1081 (9th Cir.2004). “The bias or prejudice of even a single juror is enough to violate that guarantee.” United States v. Gonzalez, 214 F.3d 1109, 1111 (9th Cir.2000). “Accordingly, ‘[t]he presence of a biased juror cannot be harmless; the error requires a new trial without a showing of actual prejudice.’ ” Id. (quoting Dyer v. Calderon, 151 F.3d 970 , 973 n. 2 (9th Cir.1998) (en banc)).

This court recognizes three forms of juror bias: (1) “actual bias, which stems from a pre-set disposition not to decide an issue impartially”; (2) “implied (or presumptive) bias, which may exist in exceptional circumstances where, for example, a prospective juror has a relationship to the crime itself or to someone involved in a trial, or has repeatedly lied about a material fact to get on the jury”; and (3) “so-called McDonough-style bias, which turns on the truthfulness of a juror’s responses on voir dire” where a truthful response “would have provided a valid basis for a challenge for cause.” Fields v. Brown, 503 F.3d 755, 766-67 (9th Cir.2007) (en banc) (citing McDonough Power Equipment, Inc. v. Greenwood, 464 U.S. 548, 554-56 , 104 S.Ct. 845 , 78 L.Ed.2d 663 (1984)). All three forms of bias potentially are implicated here, but we conclude that Olsen has not demonstrated bias under any theory.

1. Actual Bias

When an allegation of juror impartiality is raised after trial, the remedy is “a hearing in which the defendant has an opportunity to prove actual bias.” Dyer, 151 F.3d at 990 (quoting Smith v. Phillips, 455 U.S. 209, 215 , 102 S.Ct. 940 , 71 L.Ed.2d 78 (1982)). “Actual bias is, in essence, ‘bias in fact’—the existence of a state of mind that leads to an inference that the person will not act with entire impartiality.” United States v. Mitchell, 568 F.3d 1147, 1151 (9th Cir.2009) (quoting Gonzalez, 214 F.3d at 1112 ). “Actual bias is typically found when a prospective juror states that he can not be impartial, or expresses a view adverse to one party’s position and responds equivocally as to whether he could be fair and impartial despite that view.” Fields, 503 F.3d at 767 . While actual bias may be revealed by a juror’s explicit admissions, more typically it is demonstrated through circumstantial evidence. Gonzalez, 214 F.3d at 1111— 12. A defendant bears the burden of demonstrating actual bias. Martinez-Martinez, 369 F.3d at 1081 .

“The determination of whether a juror is actually biased is a question of fact,” Fields, 503 F.3d at 767 (citing Dyer, 151 F.3d at 973 ), which this court reviews “for ‘manifest error’ or abuse of discretion.” Id. (citing Gonzalez, 214 F.3d at 1112 ). “At the same time, ‘ [d]oubts regarding bias must be resolved against the juror.’ ” Martinez-Martinez, 369 F.3d at 1082 (quoting Gonzalez, 214 F.3d at 1114 ).

At the evidentiary hearing conducted in this case, the district court heard testimony from Kevin Ryan that during the summer of 2002, when he and Leavitt worked together in the same office, they had at least a dozen conversations about the Olsen case, which was in the news at the time. According to Ryan, Leavitt was “adamant” about Olsen’s guilt and would “argue the points that he read in the paper or saw in the media, the evidence that they were gathering.” Ryan testified that Leavitt was familiar with the government’s evidence regarding the internet websites that Olsen had visited, and stated: “I think he probably even told me that he’d went online and looked at one of the websites.” During these conversations, Ryan said, he challenged Leavitt about whether valid opinions could be formed before trial, explaining that “all I knew of Ken Olsen was through my wife, [that he] was a kind and gentle man.” But according to Ryan, Leavitt held to his view that Olsen was guilty and was “very adamant” that the government would be able to prove its case. Ryan testified: “It seemed like every time he’d discuss it, whatever was brought up in the media that day, was one more notch in his belt to prove guilt.” According to Ryan, other employees within their office might also have participated in these conversations, although he could not name any other particular individuals besides himself and Leavitt who had taken part.

Ryan further testified that Leavitt had a reputation within their profession, the mortgage industry, for being untruthful. According to Ryan, the owner of the company for which the two men previously worked had later indicated to Ryan that Leavitt was untruthful. Ryan said he was told similar things by a bank representative involved in the industry, and that after he and Leavitt parted ways he learned from his own bank representative that his company’s account had been “red-flagged” because of Leavitt’s association with the company. Ryan also testified that while he and Leavitt were working together “it became evident over time that he continued to lie about several different things.”

Ryan and Leavitt’s business relationship ended in the summer of 2002, after which the two men had no further contact. The conversations about Olsen’s case that Ryan recalled, therefore, took place a full year before Olsen’s trial began in July 2003.

Leavitt testified at the hearing (which took place in May 2010, seven years after Olsen’s trial) that he had no recollection of any conversations with Ryan about the Olsen case. He acknowledged that “if it was on the front headline—you know, if he said something, I may have commented back,” but he did not recall learning, in 2002, anything about a ricin investigation or anything at all about Olsen, nor that Ryan’s wife knew Olsen. Asked when he first heard of Olsen, Leavitt answered: “To the best of my knowledge, I mean, unless I read something in the paper or something, it was when I came in here ... the same day that I came into court” for jury selection.

The district court credited the testimony of both Ryan and Leavitt, and did not find their accounts to be mutually inconsistent. The court noted that the conversations Ryan described about Olsen’s case took place a full year before the trial, and that Leavitt admitted they could have taken place but simply did not recall one way or the other. The court cited a lack of any evidence that Leavitt, at the time of the trial, still held the beliefs that he expressed a year earlier or even remembered having them. It further noted that during voir dire Leavitt freely acknowledged having some familiarity with the case from the media, as did many prospective jurors, but that Leavitt assured the court that he could be impartial. In view of these considerations, the district court concluded that Leavitt’s responses on voir dire about his prior knowledge of the case were not false. Quoting Irvin v. Dowd, 366 U.S. 717, 723 , 81 S.Ct. 1639 , 6 L.Ed.2d 751 (1961), for the proposition that juror impartiality is not rebutted by the mere existence of any preconceived notion about the defendant’s guilt, the court found that Leavitt was not actually biased against Olsen.

“Whether a juror is dishonest is a question of fact[.]” Fields, 503 F.3d at 767 (citing Dyer, 151 F.3d at 973 ). Like the ultimate determination of actual bias, this finding is reviewed for clear error. Id. (citing Riley v. Payne, 352 F.3d 1313, 1317 (9th Cir.2003)). We are not firmly convinced that the district court’s findings about Leavitt’s honesty are wrong. Cer tainly, we are struck by the contrast between Ryan’s vivid recollection of having had many detailed conversations about the case and Leavitt’s own failure, a year later during voir dire, to mention these conversations or recall being exposed to anything more than “a little news blurb on one of those news break shows, something about ricin or whatever.” Ryan’s testimony about Leavitt’s reputation for untruthfulness gives us further pause. Yet we also are mindful that any casual “water cooler” conversations about the case at Leavitt and Ryan’s office, as these conversations were later characterized, would have had much less significance at the time to Leav-itt than to Ryan, whose wife had known Olsen for years and who himself had once met Olsen.

Leavitt was subpoenaed to testify at the evidentiary hearing in May 2010 with no advance warning that he was going to be questioned about his conduct during the Olsen voir dire seven years earlier, or about topical office conversations that took place a year before that. The district court was able to assess Leavitt’s demean- or as he was questioned about these matters, and it found his answers credible. See Thompson v. Keohane, 516 U.S. 99, 111 , 116 S.Ct. 457 , 133 L.Ed.2d 383 (1995) (noting that resolution of “factual issues,” including juror impartiality, “depends heavily on the trial court’s appraisal of witness credibility and demeanor”). We are not persuaded that this conclusion was manifest error or an abuse of discretion, as required to overturn a trial court’s factual findings or its conclusions about actual bias. See Fields, 503 F.3d at 767 .

Furthermore, actual bias is not proven by the mere fact that Leavitt learned about the case from the media and formed prior impressions about it a year before the trial. “[T]he Supreme Court has cautioned against presuming juror bias due to familiarity with news reports.” Crater v. Galaza, 491 F.3d 1119, 1133 (9th Cir.2007) (citing Irvin, 366 U.S. at 722-23 , 81 S.Ct. 1639 ). “It is not required [that] jurors be totally ignorant of the facts and issues involved.... To hold that the mere existence of any preconceived notion as to the guilt or innocence of an accused, without more, is sufficient to rebut the presumption of a prospective juror’s impartiality would be to establish an impossible standard.” Id. (quoting Irvin, 366 U.S. at 722-23 , 81 S.Ct. 1639 ).

2. Implied Bias

Even where actual bias has not been demonstrated, in rare instances a court will find implied bias, which is “bias conclusively presumed as a matter of law.” Mitchell, 568 F.3d at 1151 (quoting Gonzalez, 214 F.3d at 1111 ). Bias should be presumed only in “extreme” or “extraordinary” cases. Id. (citing Tinsley v. Borg, 895 F.2d 520, 527 (9th Cir.1990)). This court has recognized implied bias in only two contexts: first, “in those extreme situations ‘where the relationship between a prospective juror and some aspect of the litigation is such that it is highly unlikely that the average person could remain im partial in his deliberations under the circumstances,”’ Fields, 503 F.3d at 770 (quoting Gonzalez, 214 F.3d at 1112 ), and second, “where repeated lies in voir dire imply that the juror concealed material facts in order to secure a spot on the particular jury.” Id. (citing Dyer, 151 F.3d at 982 ). Only the latter context— repeated lies during voir dire—is implicated here. Because implied bias is a mixed question of law and fact, review is de novo. Id. (citing Gonzalez, 214 F.3d at 1112 ).

Most of this court’s decisions presuming bias as a matter of law have involved situations “where the relationship between a prospective juror and some aspect of the litigation is such that it is highly unlikely that the average person could remain impartial in his deliberations under the circumstances.” Fields, 503 F.3d at 770 . Typically the juror in question, or a close relative, “has had some personal experience that is similar or identical to the fact pattern at issue in the trial.” Gonzalez, 214 F.3d at 1112 . See, e.g., Mitchell, 568 F.3d at 1148-49 ; see also Fields, 503 F.3d at 768-770 (describing earlier cases). The theory is that even though some individuals in this position might be able to put aside their personal experiences, they nevertheless “would be lacking the quality of indifference which, along with impartiality, is the hallmark of an unbiased juror”—the court therefore “presume[s] conclusively” that these jurors will be affected in their deliberations by those experiences. Dyer, 151 F.3d at 982 . In every case where this form of implied bias has been recognized, the juror in question was not entirely forthcoming during voir dire about his or her personal experiences that were similar to the facts involved in the trial. See Fields, 503 F.3d at 773 . That is not the situation presented here.

Rather, Olsen contends that in addition to concealing his prior knowledge of the case and attitudes during voir dire, Leavitt also lied in his juror questionnaire about several matters pertaining to his own background. In Olsen’s view, these lies support a finding of implied bias under Green v. White, 232 F.3d 671 (9th Cir.2000), and Dyer v. Calderon, 151 F.3d 970 (9th Cir.1998) (en banc), the two decisions from this court that have found implied bias “not on the basis of the juror’s past history, but on the pattern of lies the juror engaged in to secure her seat on the jury.” Green, 232 F.3d at 677 .

In Dyer , a juror’s brother had been killed in a manner similar to that with which the defendant was alleged to have killed his victims, but the juror said during voir dire that no family member had ever been the victim of any crime. Dyer, 151 F.3d at 972 . This court found that the record “conclusively” showed that the juror lied repeatedly to conceal the nature of her brother’s killing, first during voir dire and later when questioned by the state trial judge about the matter. Id. at 979 . The juror also “plainly lied” when she stated during voir dire that none of her relatives had ever been accused of an offense. This was no insignificant oversight: the juror’s own husband, charged with rape, had ended up in the same jail as the defendant and had spoken with him. Id. at 973-94 . Further contradicting her voir dire responses, the juror herself had been the victim of several crimes, and “nearly every close relative of hers had been arrested” for a crime at some point. Id. at 980-81 .

In part because the district court had not made factual findings, this court did not rule on whether the juror was actually biased. Dyer, 151 F.3d at 981 . Instead, based on the “magnitude” of the juror’s lies, as demonstrated by the evidence developed in an evidentiary hearing conducted by the district court, this court drew the inference that she “lied in order to preserve her status as a juror and to secure the right to pass on Dyer’s sentence.” Id. at 982, 984 . The circumstances therefore added up to “that rare case” where juror bias must be presumed. Id. at 984 . In the course of reaching this conclusion, the court observed: “A juror ... who lies materially and repeatedly in response to legitimate inquiries about her background introduces destructive uncertainties into the process.” Id. at 983 . Even if not motivated by a personal or vindictive bias against the defendant, the court explained, such a juror is unfit to serve. Id.

These principles were later applied in Green v. White, where a juror engaged in a “pattern of misleading statements” in his juror questionnaire and during voir dire to conceal felony convictions that under state law would have disqualified him to serve as a juror. Green, 232 F.3d at 672 . Unlike in Dyer , the juror had no personal experience “similar or identical to the fact pattern at issue in the trial.” Gonzalez, 214 F.3d at 1112 . Instead, the court’s finding of implied bias was based exclusively on the “destructive uncertainties” introduced by a juror who lies “materially and repeatedly in response to legitimate inquiries about her background.” Green, 232 F.3d at 677 (quoting Dyer, 151 F.3d at 983 ).

Not surprisingly, Olsen stakes his claim of juror bias primarily on Green , attempting to draw parallels between the “pattern of misleading statements” in that case and the alleged lies by juror Leavitt here. In addition to accusing Leavitt of concealing his prior knowledge about the case, Olsen contends that Leavitt lied about his own background to get on the jury.

Leavitt provided two answers in his juror questionnaire that he later acknowledged were inaccurate when questioned about them during the evidentiary hearing. First, Leavitt indicated that he never had a lawsuit filed against him. Actually, he had been sued a number of times. At the evidentiary hearing, he was confronted with evidence that a former landlord obtained a judgment against him for unpaid rent. Leavitt testified that he never was served with papers, went to court, or paid the judgment amount, and that he first learned about the judgment years after the Olsen trial, when he had to obtain a mortgage to buy a home. Leavitt also was questioned about whether he was sued by a particular bail bond company. He testified that to the best of his knowledge, the company never actually sued him, stating: “I had a traffic violation. They bonded me out. I did not show up for court. And they came and picked me up and took me in.” Asked about two apparent judgments against him by the company, Leavitt responded: “I never went to court for it. So, if they did, they got a summary judgment. And it probably affected my credit, and I probably paid it off. And it would have been at least ten years prior [to the trial].”

Questioned about whether he was sued for writing bad checks in 1994, Leavitt answered that he “made bad mistakes” and “probably wrote bad checks,” but that he did not recall being sued and “never went to court for it.” When asked, “So let me understand. Every time you were sued you just didn’t go to court?”, he answered yes. Leavitt emphasized repeatedly that he did not recall having been served with papers to appear in court for any of those cases against him, that he fully believed he was answering the juror questionnaire honestly, and that he was “absolutely not” trying to hide any prior incidents from the court. Leavitt’s explanation for his lack of awareness about these lawsuits, as well as for the events in his life underlying them, was that when they took place in the 1990s he was in the throes of alcoholism. Later, as he put it, “I finally got married to a good woman who straightened me out,” and he thereaf ter “grew up and had children,” “became an adult,” and “put that past behind me.”

In his juror questionnaire, Leavitt also answered “no” to a question asking whether he had ever been convicted of a crime other than a traffic offense. At the evi-dentiary hearing, he explained this answer by stating that “at the time, I did not recall ever being convicted of any other crime.” Confronted with evidence that he had been prosecuted for and convicted of criminal trespass in 1994, for an incident at a bowling alley, Leavitt acknowledged the conviction and that he paid a fine for it but stated: “I honestly did not recall that event.” In explaining how he could forget such a thing, he stated: “It’s real simple. I drank an awful lot back then. I don’t now.”

The district court accepted Leavitt’s explanations for why he had forgotten about his trespass conviction and why he did not regard himself as having been sued in the other matters. It found these explanations credible, stating that “the Court does not find that he intentionally gave the Court false information” and that “mistaken, but honest juror responses” do not necessarily require a new trial. The court also observed that “there is no evidence that the inaccurate answers were an attempt to serve on a jury, let alone Mr. Olsen’s jury.” Noting its awareness that doubts regarding bias must be resolved against the juror, the court stated that it “does not hold a doubt” about Leavitt’s impartiality.

Although we review de novo the question of whether bias should be presumed, because implied bias is a mixed question of law and fact, Hamilton v. Ayers, 583 F.3d 1100, 1107 (9th Cir.2009), our review is not carried out on an entirely blank slate. Concluding that a juror’s “repeated lies in voir dire” warrant an inference of bias requires first concluding that the juror lied in voir dire. See Dyer, 151 F.3d at 979, 981 . That determination—whether a juror has been dishonest—remains “a question of fact,” about which the district court here made findings. Fields, 503 F.3d at 767 ; see Patton v. Yount, 467 U.S. 1025, 1037-39 , 104 S.Ct. 2885 , 81 L.Ed.2d 847 (1984). When this court has presumed bias from a juror’s lies, it has done so either where there were no factual findings by the trial court about the juror’s honesty or where those findings were clearly erroneous. Bias has been presumed only where the record “conclusively” showed that the juror “plainly” lied, and “no rational trier of fact could find otherwise,” Dyer, 151 F.3d at 979, 981 , or where the trial court’s “erroneous factual finding” that the juror was honest constituted “clear error ... not supported by any evidence” and the “only reasonable inference” from the facts was that the juror lied. Green, 232 F.3d at 676 .

The district court in this case credited Leavitt’s explanation that by the time of Olsen’s trial he no longer remembered having the opinions about the case that he had expressed a year earlier. As dis cussed above, we do not regard this finding as clearly wrong. We take the same view about Leavitt’s claim that the omissions in his juror questionnaire were honest mistakes. Unlike a situation where the juror’s conduct “defies an innocent explanation” and where the trial court’s finding to the contrary is “nearly inexplicable,” Dyer, 151 F.3d at 979, 981 , here it is not at all obvious from the record that Leavitt was dishonest during voir dire.

Despite Olsen’s valiant efforts to liken the circumstances here to those in Green , the comparison simply does not hold up. In Green , this court found it incontrovertible that the juror lied to conceal his criminal history, first in his juror questionnaire and then during voir dire. When called on to explain this behavior at a post-trial hearing, the juror gave explanations that were manifestly self-contradictory. Green, 232 F.3d at 672 . The juror also engaged in misconduct during the trial itself revealing his bias: it was alleged by other jurors that he said, during the proceedings, that he knew the defendant was guilty the moment he saw him and that he wished “the judge would let him go back to his place so he could get his piece” and shoot the defendant. Id. at 673-74 . When questioned about these allegations at the hearing, the juror offered “a variety of responses” that, again, were transparently self-contradictory and also conflicted with statements he had made in a declaration. Id.

Here, we think that one could fairly question the plausibility of Leavitt’s expía-nations for the omissions in his juror questionnaire. But these omissions and Leavitt’s explanations for them during the evidentiary hearing do not even approach the transparent lies of the juror in Green or the “misleading, contradictory, and outright false answers” that the juror later gave to explain those lies. Green, 232 F.3d at 678 . The “pattern of lies, inappropriate behavior, and attempts to cover up his behavior,” id. at 676 , by the juror in Green resulted in precisely the type of “extraordinary” situation where bias may be presumed as a matter of law. See Mitchell, 568 F.3d at 1151 . Leavitt’s omissions in his juror questionnaire are not as clearly intentional, nor his later explanations as transparently false, as in that case or in Dyer . It is also worth noting that these omissions cannot have been motivated by a desire to pass judgment on Olsen, because Leavitt completed and mailed in the questionnaire over two weeks before he came to court and learned in which case he might be selected to serve as a juror.

Because the facts before us, as developed in an evidentiary hearing by the district court, do not amount to the type of “extreme” situation in which manifest lies during voir dire justify a presumption of partiality, we do not find any implied bias here.

3. McDonough Bias

The third type of juror bias, “so-called McDonough-style bias,” occurs where a juror fails to answer honestly a material question on voir dire and “ ‘a correct response would have provided a valid basis for a challenge for cause.’ ” Fields, 503 F.3d at 766 -67 (quoting McDonough Power Equipment, Inc., 464 U.S. at 556 , 104 S.Ct. 845 ). The only voir dire responses from juror Leavitt that conceivably would have justified a for-cause challenge are his answers regarding his prior knowledge of and opinions about the case. We have upheld the district court’s finding that Leav-itt did not in fact answer these questions dishonestly, and accordingly no basis exists for a McDonough challenge.

Because we also have affirmed the district court’s finding that Leavitt was not actually biased and have concluded that bias cannot be presumed in the circumstances of this case, we reject Olsen’s claim that juror misconduct denied him a fair trial.

VI. Cumulative Error

Finally, Olsen contends that the cumulative effect of errors at his trial violated due process by rendering the trial fundamentally unfair. Olsen cites three alleged errors: (1) an erroneous jury instruction, (2) the prosecutor’s use of guilt-assuming hypothetical questions, and (3) the suppression of the Melnikoff-related Brady material,

Olsen did not raise the jury instruction issue on direct appeal and has not attempted to establish cause or prejudice excusing this failure; he appears to concede that the claim is procedurally barred. See United States v. Mejia-Mesa, 153 F.3d 925, 929 (9th Cir.1998). Even were it not, Olsen has given us no reason to believe that there is any merit whatsoever to this claim, which asks us to assume without any evidence that the jury adopted a strained misinterpretation of one of the court’s instructions.

With respect to the prosecutor’s use of guilt-assuming hypothetical questions, a similar version of this claim was raised and rejected on direct review. This court held that any error in permitting such questions was cured by the limiting instruction the district court later gave to the jurors, advising them to disregard the questions asked of specific witnesses that assumed the defendant possessed certain items or searched the internet for particular subjects. Olsen, 120 Fed.Appx. at 20 .

Apart from the curative effect of this limiting instruction, Olsen has not persuaded us that these allegedly guilt-assuming questions elicited answers from the witnesses that affected his substantial rights or damaged his case in any significant way. See United States v. Shwayder, 312 F.3d 1109, 1121 (9th Cir.2002) (acknowledging trial court’s error in allowing 'guilt-assuming hypothetical questions but examining whether the answers to these questions were harmful to the defense).

The district court judge, even while agreeing to give the limiting instruction, observed that in his recollection all of Olsen’s character witnesses said that their favorable opinion of him was not changed by the information with which the prosecutor confronted them. In one of the examples from the record that Olsen provides, the prosecutor asked of a character witness who had described Olsen as a peaceful person: “If somebody harmed somebody, is he a peaceful person?” The witness responded: “Depending on their intent and also I would have to say on a superficial level, that if somebody harmed somebody, they are not a peaceful person.” It is readily apparent that any damaging effect such an abstract question and response could have had would be minimal, especially since Olsen was not accused in this case of having actually harmed anyone. In the other three exam- pies Olsen provides, witnesses simply acknowledged that they did not know about the items found in Olsen’s cubicle or the internet research on poisons that he conducted.

Whether considered alone or in combination with the remaining error that Olsen cites—the government’s failure to disclose Brady material about Arnold Melnikoff— any error in allowing these questions did not render Olsen’s trial fundamentally unfair or violate his due process rights.

AFFIRMED.

. The district court conditionally dismissed the chemical weapon count prior to sentencing, by agreement of the parties based on a multiplicity issue. Olsen, 120 Fed.Appx. at 19 . . Later in the trial, Melnikoff was called back to the stand to testify that a lawsuit against him and the state of Montana by a former prisoner had just settled, with no admission of any wrongdoing. . A trial prosecutor’s speculative prediction about the likely materiality of favorable evidence, however, should not limit the disclosure of such evidence, because it is just too difficult to analyze before trial whether particular evidence ultimately will prove to be "material” after trial. Thus, "there is a significant practical difference between the pretrial decision of the prosecutor and the post-trial decision of the judge.” United States v. Agurs, 427 U.S. 97, 108 , 96 S.Ct. 2392 , 49 L.Ed.2d 342 (1976). As this court has noted, some trial courts therefore have concluded that the retrospective definition of materiality is appropriate only in the context of appellate review, and that trial prosecutors must disclose favorable information without attempting to predict whether its disclosure might affect the outcome of the trial. See Price, 566 F.3d at 913 n. 14 (noting favorably "the thoughtful analysis set forth by two district courts in this circuit” on the matter and citing United States v. Acosta, 357 F.Supp.2d 1228, 1239-40 (D.Nev.2005) ("[T]he 'materiality' standard usually associated with Brady for pretrial discovery purposes ... should not be applied to pretrial discovery of exculpatory materials.”), and United States v. Sudikoff, 36 F.Supp.2d 1196 (C.D.Cal.1999) (The standard of whether evidence would have changed the outcome "is only appropriate, and thus applicable, in the context of appellate review.... [I]t obviously cannot be applied by a trial court facing a pretrial discovery request.”)). See also United States v. Safavian, 233 F.R.D. 12, 16 (D.D.C.2005) ("The prosecutor cannot be permitted to look at the case pretrial through the end of the telescope an appellate court would use post-trial. Thus, the government must always produce any potentially exculpatory or otherwise favorable evidence without regard to how the withholding of such evidence might be viewed—with the benefit of hindsight—as affecting the outcome of the trial.”). . Rule 608(b) of the Federal Rules of Evidence authorizes courts to permit inquiry into specific instances of conduct during cross-examination if they are probative of the character for untruthfulness of the witness—subject, of course, to the balancing analysis of Rule 403. Had Olsen’s attorneys cross-examined Melnikoff about the truthfulness of his testimony in prior cases, he either would have acknowledged or denied prior questionable conduct, and counsel would have been stuck with his answers and not allowed to pursue further inquiry. See United States v. Jackson, 882 F.2d 1444, 1452 (9th Cir.1989). Nevertheless, the jury would have been able to assess his credibility when confronted and “to observe his demeanor when he answered the questions, which might have been telling.” Kohring, 637 F.3d at 905 & n. 4. . "The terms 'material' and 'prejudicial' are used interchangeably in Brady cases. Evidence is not 'material' unless it is 'prejudicial,' and not 'prejudicial' unless it is 'material.' Thus, for Brady purposes, the two terms have come to have the same meaning.” Benn, 283 F.3d at 1053 n. 9. . Although Olsen directs our attention to the places in the record where the prosecution focused on the Equate pill, examination of these passages reveals, nearly without fail, that even in these very instances the prosecution’s case focused equally on other methods of delivering ricin. To take just one example, Olsen quotes a passage from the direct examination of an FBI agent in which the prosecutor asks: “Did you find any physical evidence consistent with those delivery methods you came to learn about?” Olsen quotes the beginning of the agent’s answer ("The Equate tabs----”), but fails to note that the agent's response continued: "Also, that ricin was found in a powder form in the metal bowl, and also, that we found needles and syringes since ricin can also be injected.” . That supposition is reinforced by the fact that—according to Jeanette Ryan—Kevin Ryan even told his wife about these office conversations at the time that they took place. It was precisely because she knew that Olsen’s case had been discussed in her husband’s office a year earlier, she testified, that she was surprised to see Leavitt on the jury and immediately notified Olsen’s wife of her concem that he should not be there. With respect to Kevin Ryan's personal familiarity with Olsen, he initially told investigators that he had never met Olsen but later testified at the evidentiary hearing that his wife reminded him about how he had met Olsen at his massage therapy office, where Olsen showed Ryan a technique to use on his wife’s shoulders. . Leavitt also acknowledged that he had two or three DUI convictions—but the juror questionnaire specifically asked about convictions "other than traffic or driving offenses,” and Leavitt testified, reasonably enough, that he considered DUIs to be "traffic” offenses. . In Dyer , "the facts were not properly developed by the state court,” negating any presumption of correctness deriving from federal habeas law, and the district court did not make factual findings, despite having conducted an evidentiary hearing. This court therefore addressed the question of implied bias in the first instance, based on the record generated by the evidentiary hearing. Dyer, 151 F.3d at 979, 981 . In Green , this court found implied bias only after rejecting, as "clearly erroneous,” the state courts' factual determination that the juror did not lie about his criminal history during voir dire. Green, 232 F.3d at 678 . . The parallels that Olsen attempts to draw between his case and Green invariably crumble upon inspection. For instance, Leavitt failed to mention a misdemeanor trespass conviction that resulted in a fine; the Green juror failed to reveal a conviction for assault while he was in the Army that led him to spend six months in the brig. Green, 232 F.3d at 673, 676 . Leavitt testified that he forgot about the trespassing incident, citing his rampant alcoholism during those years; the Green juror gave no explanation for his omission, and the trial court's supposition that he forgot about it was an "unsupported assertion” not backed by "any evidence.” Id. at 676 . Significantly, the court in Green observed that if the juror in question had testified that he forgot about his assault conviction, the trial court's finding might have been reasonable, "at least enough to avoid clear error.” Green, 232 F.3d at 676 .