Opinion by Judge TALLMAN; Dissent by Judge BERZON.
*1168OPINION
TALLMAN, Circuit Judge:Petitioner-appellant Averill W. Briggs (“Briggs”) appeals the district court’s denial of his 28 U.S.C. § 2254 habeas petition challenging his jury conviction for one count of committing a forcible lewd act upon a child under 14 years of age; eight counts of aggravated sexual assault of a child under 14 years of age — including four counts of oral copulation, two counts of rape, one count of sexual penetration with a foreign object, and one count of sodomy; and first-degree burglary. Briggs is currently serving a sentence of 265 years to life for those convictions. In his petition, he argues that the prosecutor’s use of peremptory challenges to strike three African American prospective jurors violated his rights under the Equal Protection Clause of the Fourteenth Amendment. We have jurisdiction under 28 U.S.C. §§ 1291 and 2253, and we affirm.
I
A
On June 9, 2002, Briggs climbed through a second-floor window of an apartment building in the Lockwood Tevis neighborhood of Oakland, California, and sexually assaulted two 13-year-old Asian girls. In picking the jury, the prosecutor intended to rely on the following evidence on which the jury later convicted Briggs: the medical examiner’s report that confirmed one of the girls sustained oral, anal, and genital injuries from blunt penetration trauma, both victims’ pretrial identification of Briggs, and evidence of Briggs’s fingerprints recovered from the scene. There was, however, no DNA evidence.
Briggs was sentenced to 50 years to life, followed by seven consecutive sentences of 30 years to life, plus five years for a prior conviction, for a total sentence of 265 years to life.
B
The Alameda County Superior Court jury pool consisted of 65 people.1 During the selection process the district attorney used eighteen of her twenty peremptory challenges. Three of those challenges struck African American prospective jurors: Lawrence L., Georgia M., and Sam R.2 One prospective African American juror was excused for cause. The prosecutor thus struck fifteen non-African American jurors peremptorily.
After the prosecutor struck the second African American juror, Briggs challenged the action as racially motivated under Batson v. Kentucky, 476 U.S. 79, 106 S.Ct. *11691712, 90 L.Ed.2d 69 (1986), and its California analogue People v. Wheeler, 22 Cal.3d 258, 148 Cal.Rptr. 890, 583 P.2d 748 (1978). Briggs again pressed his challenge after the prosecutor used a peremptory strike to excuse the third African American prospective juror. The trial judge held a hearing at which she considered the prosecutor’s explanations for exercising her challenges as to each individual prospective juror. The judge concluded the prosecutor excused these three jurors for “non-race based reasons and they’re valid.”
After the trial, but before sentencing, Briggs moved for a new trial, based in part on the same argument that the prosecutor had misused her peremptory challenges to strike Lawrence L., Georgia M., and Sam R. Briggs also offered a comparative jury analysis to rebut the prosecutor’s race-neutral explanations for her challenges. The trial court heard argument on the motion and affirmed its previous ruling that the prosecutor had a race-neutral reason for exercising her peremptory strikes for each challenged juror:
There was a race-neutral reason for each one of the three that [defense] were concerned about. Miss M. just basically said she could not follow several points of law, there was Mr. L. who had been accused of sexual harassment, and then there was Mr. R. who basically thought teens were susceptible to coaching, and there were some other reasons for each one of them. But all the reasons that she gave, the reasons that she felt that they would not be jurors that she wanted on that panel, were legitimate.
Briggs raised his Batson challenge again on direct appeal. At the time, an open question existed in California as to whether comparative jury analysis could be considered on appeal if it was made a part of the record after the trial court ruled on the Batson motion, but at some point before judgment was entered.3 The state appellate court, nonetheless, considered the analysis in conjunction with its review of the individual jurors. The court reviewed the record to determine whether substantial evidence supported the trial court’s finding that the prosecutor’s challenges were not race based. The court then turned to the comparative analysis, finding nothing in it that “undermine[d] [its] earlier conclusion that substantial evidence supported] the trial court’s BatsonWheeler ruling” and affirmed the trial court. The California Supreme Court denied review.
Briggs filed a federal habeas petition. The district court also examined the comparative jury analysis and voir dire record and held that no Batson violation had occurred. Briggs timely appeals.
II
A Batson challenge has three steps: first, “the defendant must make a prima facie showing that a challenge was based on race”; second, the prosecution must offer a race-neutral basis for the challenge; and third, the court must determine whether the defendant has shown “purposeful discrimination.” AM v. Hickman, 584 F.3d 1174, 1180 (9th Cir.2009); see Batson, 476 U.S. at 96-98, 106 S.Ct. 1712. The sole dispute before us is whether the state appellate court reasonably applied Batson’s third step. To decide this issue, we must consider the “totality of the relevant facts” to decide “whether counsel’s race-neutral explanation for a peremptory challenge should be believed.” *1170Kesser v. Cambra, 465 F.3d 351, 359 (9th Cir.2006) (en banc) (internal quotation marks omitted).
A
“To determine whether race was a substantial motivating factor — that is, whether the defendant has shown purposeful discrimination at Batson’s third step— the trier of fact must evaluate the persuasiveness of the justifications offered by the prosecutor.” Cook v. LaMarque, 593 F.3d 810, 815 (9th Cir.2010) (internal quotation marks and brackets omitted). To decide whether the defendant has met his burden, the court must “undertake a sensitive inquiry into such circumstantial and direct evidence of intent as may be available.” Batson, 476 U.S. at 93, 106 S.Ct. 1712 (internal quotation marks omitted). This inquiry includes comparing African American panelists who were struck with those non-African American panelists who were allowed to serve. “If a prosecutor’s proffered reason for striking a black panelist applies just as well to an otherwise-similar non-black who is permitted to serve, that is evidence tending to prove purposeful discrimination to be considered at Batson’s third step.” Miller-El, 545 U.S. at 241, 125 S.Ct. 2317.
B
We review the state appellate court’s finding that the prosecutor did not engage in purposeful discrimination under the deferential standard of the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA). 28 U.S.C. § 2254(d)(2). Under § 2254(d)(2), we must defer to the California court’s conclusion that there was no discrimination unless that conclusion “was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.”4 Here our standard is doubly deferential: unless the state appellate court was objectively unreasonable in concluding that a trial court’s credibility determination was supported by substantial evidence, we must uphold it. See Rice v. Collins, 546 U.S. 333, 338-42, 126 S.Ct. 969, 163 L.Ed.2d 824 (2006); see id. at 341-42, 126 S.Ct. 969 (“Reasonable minds reviewing the record might disagree about the prosecutor’s credibility, but on habeas review that does not suffice to supersede the trial court’s credibility determination.”).
Although the dissent ultimately “reeite[s] the proper standard of review, ... [it] improperly substitute^] its [de novo] evaluation of the record for that of the state [appellate] court.” Rice, 546 U.S. at 337-38, 126 S.Ct. 969 (overturning the Ninth Circuit). While citing AEDPA, the dissent repeatedly suggests that we cannot credit the prosecutor’s justifications because the defense’s characterization of the challenged jurors’ questionnaire answers contradict the prosecutor’s characterization at the hearings before the trial judge. Dissent at 1184-85, 1186, 1188 n. 6. The dissent seems to conclude that because we cannot independently verify the answers from the questionnaires as they are not in the record, the defense’s characterization is equally, -if not more, plausible despite the state court determinations to the contrary. However, “AEDPA imposes a highly deferential standard for evaluating state-court rulings and demands that state-court decisions be given the benefit of the doubt,” Felkner v. Jackson, — U.S. -, 131 S.Ct. 1305, 1307, 179 L.Ed.2d 374 (2011) (per curiam) (internal quotation marks omitted) (overturning the Ninth Circuit). The dissent’s readiness to doubt the state court determination based on the *1171defendant’s characterization of the record does not apply the appropriate level of deference Congress and the United States Supreme Court have required of us.
Additionally, it is widely acknowledged that the trial judge is in the best position to evaluate the credibility of the prosecutor’s proffered justifications. See e.g., Rice, 546 U.S. at 343, 126 S.Ct. 969 (Breyer J. concurring) (“[T]he trial judge is best placed to determine whether, in a borderline case, a prosecutor’s hesitation or contradiction reflect (a) deception, or (b) the difficulty of providing a rational reason for an instinctive decision.”); Hernandez v. New York, 500 U.S. 352, 111 S.Ct. 1859, 114 L.Ed.2d 395 (1991) (“As with the state of mind of a juror, evaluation of the prosecutor’s state of mind based on demeanor and credibility lies peculiarly within a trial judge’s province.” (internal quotation marks and citation omitted)). Given that the trial court did have the benefit of viewing the questionnaires and the prospective jurors who answered them when making a determination under Batson, and that the California Court of Appeal presumably had those questionnaires on review, we must defer to these credibility and factual findings.5 See Rice, 546 U.S. at 338-39, 126 S.Ct. 969 (“[A] federal habeas court can only grant Collins’ petition if it was unreasonable to credit the prosecutor’s race-neutral explanations for the Batson challenge. State-court factual findings, moreover, are presumed correct; the petitioner has the burden of rebutting the presumption by ‘clear and convincing evidence.’ ”) (quoting Miller-El, 545 U.S. at 240, 125 S.Ct. 2317). Thus it would be anathema to AEDPA if we were to assume that the petitioner’s contentions about the questionnaires are true simply because the record before us does not contain the excused jurors’ questionnaires. The burden to disprove the factual findings rests with Briggs. 28 U.S.C. § 2254(e)(1) (requiring “clear and convincing evidence” to rebut “a determination of a factual issue made by a State court”).
C
The trial judge held two hearings on Briggs’s Batson motion before concluding that the prosecutor’s explanations were not pretextual. The California Court of Appeal carefully reviewed the record for substantial evidence in upholding the trial court’s findings, and the state appellate court’s determination on review is “entitled to appropriate deference.” Cook, 593 F.3d at 815; see Felkner, 131 S.Ct. at 1307.
The state appellate court also considered Briggs’s comparative analysis and found it unpersuasive. The court rejected Briggs’s analysis because it was incomplete in that it relied primarily upon comparisons of the jurors’ questionnaire answers and failed to account for the differences between the same jurors’ answers during voir dire. Thus a careful examination of the full record, along with consideration of the prosecutor’s justifications as a whole, supported the trial court’s determination that the challenges were non-discriminatory. Moreover, on federal habeas review, the district court evaluated the voir dire transcript, conducted comparative juror analysis where possible, and found that the state-court’s determination was not objectively unreasonable. We agree.6
*11721. Juror Lawrence L.
The prosecutor used her first peremptory challenge to strike Lawrence L. As justification for the challenge she offered five reasons:' (1) Lawrence L. had been accused of sexual harassment, which he failed to include on his questionnaire; (2) he stated that it would be difficult for him to convict on the word of only one witness because of his own experience as the subject of a sexual-harassment investigation; (3) he admitted to the court that he would hold the prosecution to a higher standard of proof than required by law; (4) he thought that the believability of teenagers was affected by what they see and hear at home; and (5) he failed to answer other written questions that pertained to the burden of proof or the types of witnesses who would be presented at trial.
Both the trial court and the appellate court concluded that the prosecutor’s strike rested primarily on Lawrence L.’s involvement in the workplace-sexual-harassment investigation and the concern that his involvement would affect how he viewed the witnesses in Briggs’s case: two teenage victims of sexual assault.
The state appellate court found that both Lawrence L.’s questionnaire and his statements during voir dire supported the trial court’s determination that the prosecutor’s reasons for striking Lawrence L. were race-neutral. For example, during voir dire the following exchange between Lawrence L. and the trial judge occurred:
Q. [prosecutor] Based on your experience, though, do you think you would be more demanding than the law requires in terms of the evidence?
A. Yes.
Q. [Court] Now, everyone, and we’ve talked about this, the requirement is proof beyond a reasonable doubt. A. Yes.
Q. Not proof beyond any possible doubt. So her standard is proof beyond a reasonable doubt. Would you hold her to proof beyond any reasonable doubt?
A. Yes.
At another point during voir dire, Lawrence L. also indicated that it would be difficult for him to be a fair juror because he was “in a sexual harassment case, and ... didn’t like the way it came out.” No other juror was accused of sexual harassment, so comparative analysis is of little help. See Cook, 593 F.3d at 817 (comparison to a juror who is not “otherwise similar” nullifies the comparative value). Based upon our review of the record, it was not objectively unreasonable for the California Court of Appeal to find that substantial evidence supported the trial court’s determination that the prosecutor challenged Lawrence L. for race-neutral, legitimate reasons.
2. Juror Georgia M.
The prosecutor exercised her fourth challenge to excuse Georgia M. The *1173prosecutor gave the following justification for challenging Georgia M.:
Mr. Steckler, when he got up to do his voir dire of Miss M. with some of the other jurors, directed a question, first a comment then a question to Miss. M. He said to Miss M. as well as [Lawrence L. and two other jurors], he said, it sounded as though when you were talking to the prosecutor, that you were not going to follow the law at the end of the case, and he then moved, ... which signifies to me an acknowledgment on the part of the defense that there were answers given that warranted that remark by counsel that he made as a blanket remark to four jurors, two of which were African American.7
He then went on to ask about the standard of proof, beyond a reasonable doubt, and Miss M. said, in fact, volunteered and said, well, if there’s a slight doubt in your mind, then there’s a reasonable doubt.
That is exactly why I was concerned for Miss. M., and that’s corroborated by what she said to me when I probed. And she said that in response to my question: would you need a little bit more evidence because this is a sex case? She said yes, I do need more evidence. Other jurors did not have a question with that area that are seated.
Now, I should point out that Miss M. also indicated on page 18 of her questionnaire, as did Mr. S., who was not African American, and Mr. H., who was not African American, that I kicked, that they would hesitate to convict on the word of one witness alone.
She also indicated that whether sex victims were more or less believable, that she did not have an opinion.
Now, her ambivalence was significant to me because that, compounded with her lack of clear responses, indicated that she was impressionable and was impressionable in the direction of requiring more of me. And that’s further corroborated by, when asked, do you think you’d require DNA, she said, depends how strong the other evidence is.
So this is someone who is clearly looking for, at least how I felt, stronger evidence than I would otherwise be required to present.
*1174Now, she also said that she’s not a good judge of telling the truth. And when I asked her, would you be hesitant to convict a defendant if you believed the victim’s testimony beyond a reasonable doubt, the answer there, memorialized as well on page 18, is I don’t know. Again, with regard to sentence, her answer is “I don’t know.”
Further, I did not have a good rapport with her. I did not get a warm feeling from her. I actually got a cold stare with little eye contact, had no connection with her. And I noted. that there was actually good rapport between defense attorney and herself.
Her answers were not answers that gave me any comfort, and the jurors that were coming up in the box were much stronger for me than she was. I felt like I would have to do more than the law required to persuade her, and I think that’s corroborated again to go— going back where I started, the defense attorney’s remark to her that it sounded as though you were not going to follow the law at the end of the case.
And, clearly, somebody who thinks slight doubt equals reasonable doubt is a scary juror for me.
The prosecutor’s fundamental concern was that Georgia M. would hold the prosecution to a higher burden of proof than the law required. The prosecutor offered several examples of Georgia M.’s questionnaire answers or voir dire statements that, cumulatively, supported this concern. Each detail the prosecutor cites may not necessarily constitute a stand-alone justification, but in total provided support for her overall concern with this juror. See Cook, 598 F.Bd at 819-20 (noting individual factors that contributed to the prosecutor’s concern with the “juror’s overall demean- or”). The appellate court affirmed the trial court’s determination that the prosecutor’s overall concern regarding the burden of proof was credible. Six out of the seven factors identified by the prosecutor contributed to or compounded this concern.8 We consider each related reason that the prosecutor cites for Georgia M. in light of that overarching concern.9
*1175First, and most significantly, the prosecutor pointed out that not only did Georgia M. equate “slight doubt” with “reasonable doubt,” but she also opined that the burden of proof increases with the seriousness of the crime charged.10 The voir dire transcript and questionnaires support the prosecutor’s justification. No other accepted juror made a similar combination of statements that would create doubt about that juror’s ability to faithfully apply the law. In particular, no seated juror equated reasonable doubt with slight doubt or volunteered any similar statements.
As to the second justification — Georgia M.’s statement that a rape case would require more evidence than an auto-theft case — Jurors 6, 8, and 10 also answered “yes” to the question, “Would you require more evidence in a sexual assault case as opposed to another type of crime such as auto theft?” But none of those jurors repeated that answer during voir dire. In fact, each immediately answered “no”— more evidence is not needed — -when asked. In contrast, Georgia M. was asked three times, twice by the prosecutor and once by the court, before she retracted her statement regarding the burden of proof. The prosecutor is not required to ignore Georgia M.’s repeated “yes” answer simply because she eventually acquiesced to the judge’s explanations. Cf. Rice, 546 U.S. at 341, 126 S.Ct. 969(“That the prosecutor claimed to hold concerns despite Juror 16’s voir dire averments does not establish that she offered pretext.”); Cook, 593 F.3d at 820 (crediting prosecutor giving more weight to initial questionnaire answers than voir dire answers when exercising challenges).
We also find support for this justification from the prosecutor’s use of a peremptory strike to excuse at least one non-African American potential juror who indicated on her questionnaire, and reaffirmed during voir dire, that she would hold the prosecution to a higher standard of proof, even though she retracted that statement when pressed by the court.11 See Rice, 546 U.S. at 341, 126 S.Ct. 969(“Even if the prosecutor was overly cautious in this regard, her wariness of the young and the rootless could be seen as race neutral, for she used a peremptory strike on a white male juror, Juror 6, with the same characteristics.”); Ngo v. Giurbino, 651 F.3d 1112, 1116-17 (9th Cir.2011) (finding support for determination that prosecutor’s justifications were not pretextual where prosecutor also struck other prospective jurors who presented similar characteristics).
The prosecutor’s third justification is also closely related to her overall concern that Georgia M. would hold the prosecution to a higher burden. First, Georgia M. suggested that she would hesitate to convict on the word of one witness alone. *1176Two seated jurors, 1 and 10, had similar answers to that question. Juror 1 simply checked “yes”, whereas Juror 10 said “maybe” depending upon the factual circumstances. When questioned on voir dire, Juror 1 answered that the word of one witness would be sufficient, as did Juror 10. The prosecutor’s justification is somewhat weakened by Georgia M.’s similar answer on voir dire, that she would be able to convict on the word of one witness if she believed that witness. Careful examination of the record shows, though, that neither Juror 1 nor 10 expressed the same ambivalence or lack of understanding that Georgia M. exhibited, which was what the prosecutor identified as troubling.
Instead, Juror 1 was able to clearly communicate and clarify his questionnaire answers. Juror 10 also gave succinct and direct answers to the questions posed during voir dire. Georgia M., on the other hand, expressed more than once that she did not understand various questions or legal concepts, and she agreed to follow the law only after the court interjected to explain certain points of law. The prosecutor explained that this ambivalence led her to believe that Georgia M. could be influenced by other jurors in how she would apply the burden of proof. As the prosecutor noted, she also struck two non-African American jurors for the same reason, lending further support to the finding that the reason for the strike was not pretextual. Ngo, 651 F.3d at 1116. Briggs has not met his burden under AEDPA: nothing in the record dictates a finding that this reason was clearly pretextual.
Viewed in isolation, the prosecutor’s fourth justification is somewhat weak. Review of the record, however, supports the state court’s finding that this reason was not pretextual. Georgia M. answered that she did not know if sexual assault victims were more or less believable than other victims. And it was her uncertainty that concerned the prosecution: “Her ambivalence was significant to me because that, compounded with her lack of clear responses, indicated that she was impressionable .... ” Each of the twelve seated jurors and the four alternates either wrote “no” without qualification or indicated that believability is based upon the individual. Although, as the dissent points out, a “no” answer to this question can be read to mean that the person has no opinion, thus indicating ambivalence, this does not make the prosecutor’s concern that Georgia M. might have been easily influenced by other jurors clearly pretextual as it is the complete picture that we must analyze. Here, the prosecutor identified several reasons that led her to believe that Georgia M. would be easily influenced by fellow jurors. The fact that seated jurors shared one of those characteristics does not ineluctably lead to the conclusion that the prosecutor’s concern was pretextual.
The prosecutor’s fifth justification is also weak, but nonetheless supported by comparison to the seated jurors. The prosecutor cited Georgia M.’s “it depends” answer to the question whether, in the absence of DNA evidence, she could convict a defendant of a sexual crime. While eight other jurors gave similar answers to the DNA question, not one of these jurors shares all of the troubling characteristics that the prosecutor identified as contributing to her doubt of Georgia M.’s ability to apply the appropriate burden of proof. During both hearings on the Batson motion, the prosecutor explained that her primary concern was compounded by Georgia M.’s ambivalence about whether she would require DNA evidence. The trial judge, after having observed the entire jury selection process, credited the prosecutor’s explanation, and the appellate court affirmed. Nothing in the record contradicts those determinations.
*1177The prosecutor’s sixth justification — that Georgia M. said she was not good at assessing who is telling the truth — plausibly could compound the prosecutor’s concern that Georgia M. would not be a good juror because she could be unduly influenced by her fellow jurors. Juror 8 said that she was “not really” a good judge of whether someone was telling the truth and Juror 1 said that he was “usually” a good judge of telling the truth, but not always. These similar answers somewhat undermine the prosecutor’s reasoning, but the record does not rise to the level of clear pretext considering the combined effect of Georgia M.’s questionnaire and voir dire answers.
Finally, the prosecutor noted that she did not have a “good rapport” with Georgia M. and that Georgia M. gave her a cold stare, while Georgia M. seemed to have a good rapport with the defense counsel. Although it could be, as the dissent points out, difficult to evaluate a prosecutor’s reason if she stated only that she had a bad “feeling,” this is simply not the case here.12 The state trial court did not make a specific finding about this justification, thus we cannot presume that the trial court credited or discredited this reason, but instead base our determination upon the other justifications that the prosecutor offered. See Snyder, 552 U.S. at 479, 128 S.Ct. 1203. Consequently, because AEDPA “ ‘demands that state-court decisions be given the benefit of the doubt,’ ” we cannot say that the record shows it was objectively unreasonable for the trial judge to credit the prosecutor’s justification as a whole. Cullen v. Pinholster, — U.S.-, 131 S.Ct. 1388, 1398, 179 L.Ed.2d 557 (2011) (quoting Woodford v. Visciotti 537 U.S. 19, 24, 123 S.Ct. 357, 154 L.Ed.2d 279 (2002) (per curiam)); see also Rice, 546 U.S. at 341, 126 S.Ct. 969.13
The state court noted that the prosecutor’s primary justifications for Georgia M. were her misunderstanding of reasonable doubt and her opinion that the burden of proof should be higher in a rape case. Four of the prosecutor’s secondary justifications also support the prosecutor’s primary concern. Viewing the justifications together, the most generous reading *1178would suggest only that the state court had reason to question the prosecutor’s justification for the strike of Georgia M. Rice, 546 U.S. at 341, 126 S.Ct. 969. “That does not, however, compel the conclusion that the [state] court had no permissible alternative but to reject the prosecutor’s race-neutral justifications and conclude [petitioner has] shown a Batson challenge.” Id. We cannot say, applying AED-PA, that the California Court of Appeal was objectively unreasonable in finding that substantial evidence supported the trial court’s determination that the prosecutor exercised her peremptory challenge for race-neutral reasons.
3. Juror Sam R.
The prosecutor offered the following reasons for challenging Juror Sam R.: (1) his demeanor and manner of responding to the prosecutor’s questions on voir dire suggested that Sam R. was not taking the selection process seriously; (2) Sam R. was flippant and evasive in his answers; (3) Sam R. did not feel that he needed to talk to his teenage daughters about the potential for sexual assault; (4) Sam R. indicated during voir dire that he thought teenagers were more susceptible to coaching; (5) Sam R.’s answer that sexual assault victims are sometimes less believable because of age or personal background; (6) Sam R.’s “yes” answer to the question of whether he had a bias and his failure to explain that answer on voir dire; and (7) the prosecutor’s perception that Sam R. would look for physical evidence because of his questionnaire answers.
The district court noted that the prosecutor’s fundamental concerns were her first and second justifications: Sam R.’s offhand demeanor, as well as his curt and sharp answers to her questions. Here, the trial judge was in the best position to evaluate the credibility of the prosecutor’s demeanor-based reasons — the California Court of Appeal deferred to that evaluation, and we must as well. “[R]ace-neutral reasons for peremptory challenges often invoke a juror’s demeanor ..., making the trial court’s firsthand observations of even greater importance.” Snyder, 552 U.S. at 477, 128 S.Ct. 1203. The prosecutor clearly articulated how Sam R.’s behavior influenced her perception that he was not taking the process seriously and therefore would not be a good juror. This reason is credible. Indeed, the record shows that Sam R.’s answers were short and often vague or evasive. Our review of the record reveals at least four exchanges between the prosecutor and Sam R. during which he answered her questions with questions or avoided giving any direct answer. Cf. McClain v. Prunty, 217 F.3d 1209, 1223 (9th Cir.2000) (finding a demeanor-based challenge pretextual where “the prosecutor did not explain the significance of [the juror’s body language] or otherwise indicate how that gesture evidenced bias”). Thus, npthing in the record shows that this reason was clearly pretextual.
The prosecutor’s third justification is also not clearly pretextual. It seems to us completely logical that a prosecutor in a case that charges horrific sexual crimes against young teenage girls would be concerned with a juror who responds that he “never found the need to” discuss sexual assault with his own daughters when they were growing up. Briggs points to no seated juror who expressed the same ambivalence, nor can we find anything in the record to support his claim that this reason was clearly pretextual.
The prosecutor’s fourth reason is also supported by the record. During voir dire the prosecutor asked Sam R.: “Well, do you think that teenagers are more prone to being coached [as witnesses]?” Sam R.’s response was: “At times, they are.” We agree with both Briggs and the dissent *1179that there was some confusion related to this justification and the prosecutor’s next justification. “Seizing on what can plausibly be viewed as an innocent transposition makes little headway toward the conclusion that the prosecutor’s explanation was clearly not credible!]]” however. Rice, 546 U.S. at 340, 126 S.Ct. 969. It is plausible that the confusion merely stems from the related nature of the two questions,14 and “[i]t is a tenuous inference to say that an accidental reference with respect to one” of Sam R.’s answers undermines the prosecutor’s credibility. Id.
By including these questions on the questionnaire and focusing on them during voir dire, the prosecutor was presumably attempting to ascertain whether the potential jurors would find either sexual assault victims or teenagers or both less credible as witnesses. Her concern clearly related to the fact that the victims — teenagers subject to a violent sexual assault — possessed those characteristics and would be testifying. The prosecutor’s reason is further supported by her statement that she believed part of the defense strategy would be to argue that the victims’ identifications of Briggs were coached or suggested to them. And, indeed, Sam R. during voir dire said that teenagers could be coached. Thus we cannot, under AEDPA, simply credit the defense’s version of Sam R.’s questionnaire answer. Even if we were discussing the same question in the fourth and fifth justifications — which we are not — we must credit the trial court’s determination.
From the transcript it is apparent that the trial judge had the questionnaire in front of her during the Batson challenge. The trial judge recognized the confusion and asked the prosecutor to identify the exact question to which she was referring. The prosecutor gave the judge a page and question number, at which point the proceedings continued. The reasonable inference is that the answers matched up with the prosecutor’s justification, not that the trial court eschewed the constitutional standards set forth in Batson by crediting a demonstrably false and pretextual justification. See Visciotti, 537 U.S. at 24, 123 S.Ct. 357 (“Th[e] readiness to attribute error [to the state court] is inconsistent with the presumption that state courts know and follow the law.”).
We also find support for this justification by comparing Sam R. to other jurors whom the prosecutor also challenged: three non-African American jurors were excused who voiced similar opinions about the susceptibility of teenagers to coaching as witnesses. Mr. S. stated on his questionnaire, and then confirmed during voir dire, that he believed teenagers were more susceptible to outside influence. Mr. H. also answered on his questionnaire that he believed teenagers might “just say what an adult has told” them. He explained during voir dire that it may depend on the individual, but that “it’s more so with young teenagers.” Ms. B. also agreed that teenagers would be “easier to lead” as witnesses. These comparisons support the conclusion that this reason was not pretextual.
The fifth justification was Sam R.’s questionnaire answer that sexual assault victims could be less believable due to age or personal background and his failure during voir dire to explain this opinion adequately. The prosecutor attempted to *1180clarify Sam R.’s answer, but instead of directly answering her queries he equivocated:
Q. You said that sometimes you think cases with victims of sexual assault are sometimes less believable, sometimes they’re less believable because of age and personal background. What do you mean by that?
A. Did I say that?
Q. Yeah. I put a big old circle around it.
A. Maybe I misunderstood the question.
Q. Well the question was, do you have any opinions [that] victims of sexual assault are more or less believable. And that was the gist of the question. And then you thought that sometimes they’re less believable. Why do you say that?
A. I don’t. Like I say, I misinterpreted the question, so.
Q. ... What is it about somebody’s age or personal background, in your view, that could affect the believability of a sexual assault victim?
A. They could be coached in the questions, but I don’t think they’re less believable.
Neither Briggs nor the dissent addresses this justification or offers a comparative analysis on this point. A review of the record, however, supports a finding that this was a non-race-based justification. As we previously noted, the prosecutor was explicit about her concern that Sam R. would not find the victims credible as witnesses. This exchange supports that reasoning. Nothing in the record shows that it was clearly pretextual.
As the sixth reason, the prosecutor cited Sam R.’s questionnaire answer of “yes” to the question whether he had a bias. No other seated juror answered “yes” to this question on the questionnaire. Thus, comparative analysis is of little value. The prosecutor, however, also cited the voir dire exchange on this question as further evidence of Sam R.’s evasiveness. When confronted with his questionnaire answer, he again replied with a question: “Do I have a bias? ... I probably misunderstood your question.” When pressed he answered “no,” he did not have a bias. At a minimum, Sam R. was not forthcoming with his answers during voir dire. This factor further supports the appellate court’s conclusion that substantial evidence supported the trial court’s determination that Sam R.’s manner during voir dire was a non-race-based justification.
Finally, the prosecutor cited several of Sam R.’s questionnaire answers and responses during voir dire for her perception that Sam R. was a “physical evidence” type of guy. The California Court of Appeal found that the record did not support this justification. The appellate court, however, also found that the prosecutor’s main concern was Sam R.’s off-hand demeanor, and the court concluded that the existence of one weak justification did not prove purposeful discrimination or require reversal of the trial court’s determination. Rice, 546 U.S. at 340-41, 126 S.Ct. 969 (“Concerned about the constitutionality of such a strike, the trial court made clear that it would not accept gender as a race-neutral explanation.... The prosecutor provided a number of other permissible and plausible race-neutral reasons, and Collins provides no argument why this portion of the colloquy demonstrates that a reasonable factfinder must conclude the prosecutor lied about the eye rolling and struck Juror 16 based on her race.”); see also Cook, 593 F.3d at 819 (concluding that the state court’s determination that the peremptory strike was not substantially motivated by race was not objectively unreasonable even where “the prosecutor *1181gave four legitimate and two illegitimate grounds for striking [the juror]. The prosecutor’s two primary motivations are quite persuasive and are unrefuted by the record”). As the record does not refute the prosecutor’s main concern with Sam R., the appellate court’s conclusion that valid grounds — not race — motivated the strike was not objectively unreasonable.
4. Cumulative Evidence
The prosecutor struck three African American jurors. There is no dispute that this fact calls for a searching inquiry. A close review of the record shows, though, that the state court was not objectively unreasonable in finding that the three jurors were excused for race-neutral reasons.
When viewed as a whole, the record reveals that the prosecutor consistently questioned jurors in the same vein as her main concerns with the three African American jurors. For example, the record reveals that the prosecutor questioned almost every juror about whether he or she would require DNA to convict, especially when the individual had given anything short of a “no” answer on the questionnaire. This concern apparently stemmed from the prosecution’s lack of DNA evidence. Similarly, she consistently questioned jurors about the burden of proof in a rape case and whether each juror could convict on the word of one witness alone— a line of questioning linked to her anticipation that the victims would be the only testifying eye witnesses for the prosecution at trial. The consistency in her lines of questioning to jurors of all races and its relevance to the circumstances support crediting the prosecutor’s fundamental concerns with Lawrence L., Georgia M., and Sam R. as potential jurors.
III
The trial court credited the prosecutor’s justifications, and the California Court of Appeal found that substantial evidence supported the determination. Both the district court’s and our own review of the record fail to show purposeful discrimination on the part of the prosecutor. Although some of the prosecutor’s justifications may be weak when dissected and examined individually, the central justifications for each juror are sound and permissible. Under AEDPA’s deferential standard of review, we cannot conclude that the California Court of Appeal’s finding— that there was no racial discrimination— was objectively unreasonable.
We therefore conclude that Briggs did not suffer any violation of his rights under the Fourteenth Amendment.
AFFIRMED.
'. Briggs's opening brief makes much of the fact that this trial was held in Oakland, California, without any African American jurors. By focusing solely on Oakland, however, Briggs overlooks the fact that the jury pool is drawn from the entire county of Alameda. Thus, instead of drawing from a population solely from Oakland, which the petitioner cites as being 30.8% African American, the pool was drawn from a county that in 2000 was 14.9% African American and in 2010 was 12.6 % African American. U.S. Census Bureau, Alameda County — General Demographic Characteristics: 2000, http://factfinder2. census.gov/faces/tableservices/jsPpages/ productview.xhtml?src=bkmk (last visited Apr. 13, 2012); U.S. Census Bureau, Alameda County, California: 2010 (Jan. 31, 2012), http://quickfacts. census. gov/qfd/states/0 6/ 06001.html. In any event, this point is misleading because “under a Batson challenge, we do not hold against the government the fact that the panel lacked African-American members.’’ United States v. Collins, 551 F.3d 914, 920 (9th Cir.2009). Briggs has never asserted that the potential jurors were drawn from a non-representative cross-section of the community. See Duren v. Missouri, 439 U.S. 357, 363-64, 99 S.Ct. 664, 58 L.Ed.2d 579 (1979).
. Throughout this opinion we use juror numbers or abbreviated names instead of the jurors’ full names to protect their privacy.
. The Supreme Court later answered this question in the affirmative. Miller-El v. Dretke, 545 U.S. 231, 241 nn. 1, 2, 125 S.Ct. 2317, 162 L.Ed.2d 196 (2005) (using a comparative analysis that relied upon the voir dire transcript that was a part of the record before the state court, where the defendant offered the analysis after judgment).
. We apply § 2254(d)(2) instead of § 2254(e)(1), because the evidence the petitioner relies upon is found in the record that was before the California Court of Appeal. See Kesser, 465 F.3d at 358.
. Indeed, during the colloquy on Sam R., some confusion arose as to which questionnaire and question the prosecutor was referring. The trial court asked for clarification, after which the court allowed the hearing to continue, leading to the reasonable inference that the trial court was simultaneously reviewing the challenged jurors' questionnaires during the hearing and found no discrepancy. See infra pages 1179-80.
. Where the state court conducted comparative analysis and determined that the prosecu*1172tor did not exercise her peremptory challenges in a discriminatory manner, AEDPA deference applies and we need not undertake comparative analysis de novo. The state court was explicit that it considered and rejected Briggs's comparative analysis, but it did not give a detailed explanation for rejection of each of the proffered justifications. In Miller-El, 545 U.S. at 241, 125 S.Ct. 2317, the Court “presumed [that] the trial court and state appellate court did not undertake [such] analysis because [it] was not detailed in their opinions.” Green v. LaMarque, 532 F.3d 1028, 1030 n. 2 (9th Cir.2008). Here the state appellate court did, however, give some specific reasons why the comparative analysis failed to show purposeful discrimination at step three. Thus, we include a detailed comparative analysis only where appropriate.
.' As a preliminary matter, we note that the dissent finds this rationale "false” and therefore demonstrably pretextual, dissent at 1184— 85, because defense counsel addressed three jurors, none of whom was Georgia M. There were, however, two instances during voir dire when defense counsel addressed this same group of jurors. The first encounter follows:
[Defense counsel] I have one kind of a mini group question, and this is primarily for Miss H., Mr. C., Mr. [Lawrence] L. and somewhat a little bit for Ms. [Georgia] M., and this is concerning the concept of proof beyond a reasonable doubt.
Just a show of hands who knows what proof beyond a reasonable doubt means.
PROSPECTIVE JUROR [GEORGIA M.]: I don’t know if I .know what it means____I think I know what it means. It’s when you — when a case or evidence or whatever, if there’s a slight doubt that in your mind— okay.
In the other instance, defense counsel directed another group question pertaining to the burden of proof at Miss H., Mr. L., and Mr. C. and then immediately went on to question Georgia M. on a different topic. It appears that the prosecutor's reference confused the exact wording of the question that defense counsel posed in the first instance with the wording in the second, but her worry was explicitly identified as Georgia M.'s answer that "slight doubt” was equivalent to "reasonable doubt.” Accordingly, this reason is not so easily labeled "false” and cannot be rejected out of hand. See Rice, 546 U.S. at 340, 126 S.Ct. 969 ("Seizing on what can plausibly be viewed as an innocent transposition makes little headway toward the conclusion that the prosecutor’s explanation was clearly not credible.”).
. Although Briggs identifies eight justifications, a careful reading of the record reveals that, when the prosecutor referred to Georgia M.’s answer on page 18 of the questionnaire to whether she could convict on the word of one witness, and then referred to Georgia M.'s answer to whether she could convict on only the victim’s uncorroborated testimony (also on page 18), the prosecutor was actually referring to the same questionnaire question. The only question on page 18 to which the prosecutor could have been referring was:
The law does not require a victim’s testimony to be corroborated by other evidence. That is, if you believe a victim’s testimony beyond a reasonable doubt, that alone is sufficient to find a defendant guilty.
a. Would you at all hesitate to convict a defendant of the charges if you believed a victim's testimony beyond a reasonable doubt?
Yes_No__
b. Would you at all hesitate to acquit a defendant of the charges if you disbelieved a victim’s testimony beyond a reasonable doubt?
Yes_No_
Thus we treat this as one justification and not two separate ones.
. Contrary to Briggs's assertion that the sheer number of justifications belies pretext, upon careful examination it becomes evident that many of the prosecutor's justifications were facets of a deeper underlying concern that Georgia M. would not apply the correct legal standard. The quantity of the prosecutor's justifications alone, without examination of the quality of those justifications, cannot prove purposeful discrimination. See Rice, 546 U.S. at 340-41, 126 S.Ct. 969 (finding state-court determination that prosecutor did not exercise his peremptory challenges in a discriminatory manner was not objectively unreasonable despite concern about constitutionality of one justification where "[t]he prosecutor provided a number of other permissible and plausible race-neutral reasons”).
. We recognize that the concept of reasonable doubt can be difficult to explain and, alone, this justification may not support the use of the strike. However, the prosecutor considered the compound effect of Georgia M.’s statement and the other answers that Georgia M. gave. The trial court credited that reasoning, the state court of appeal affirmed it, and — unless objectively unreasonable — we must defer to these determinations under AEDPA.
. The dissent takes issue with our comparison of Georgia M. with this other non-African American excused juror, proposing alternative reasons why this juror was excused. See dissent at 1184-85. Because the prosecutor was not asked to explain her peremptory strike of this juror, it is speculation why that juror was excused. We note only that she presented similar characteristics to Georgia M. The Supreme Court in Rice, 546 U.S. at 341, 126 S.Ct. 969 used this very technique to find support for the prosecutor’s justification for the use of a peremptory without comment on or consideration of other potential reasons the prosecutor excused the non-black potential juror.
. The dissent’s citation to United States v. Horsley, 864 F.2d 1543 (11th Cir.1989) (per curiam), is inapposite. We note, first, that a less deferential standard of review applied there, as Horsley was a direct appeal. Second, the prosecutor’s justification in Horsley for excusing an African American juror was far different than the prosecutor’s observation of rapport here. In Horsley, the prosecutor simply said, "I don't have any particular reason. I just got a feeling about him as I have about Mr. Gonzalez, and several others.” Id. at 1544. Having no articulable reason is a far cry from the prosecutor’s detailed justification for excusing Georgia M., in which rapport played a minor role. Furthermore, we note that we could not find, and the dissent does not cite, any Ninth Circuit precedent to support the distinction between a "rapport” and a demeanor-based justification.
. The dissent argues that we must reject the rapport justification simply as it is unfalsifiable or non-verifiable on the record because it would be too easy for prosecutors to mask racial animus by claiming a lack of rapport with a juror. At its core this argument is not new or novel, and in fact similar contentions have been rejected. In Batson itself Justice Marshall noted the inherent difficulty in evaluating a prosecutor’s justifications, 476 U.S. at 106, 106 S.Ct. 1712 (Marshall, J., concurring). And in Rice, 546 U.S. at 343, 126 S.Ct. 969 (Breyer, J., concurring), Justice Breyer echoed this concern asking: "Insofar as Bat-son asks prosecutors to explain the unexplainable, how can it succeed?” Given the long history of Batson and its progeny, the Supreme Court has grappled with these questions many times, but it never has changed the framework of Batson. As such, even though we recognize the inherent problem of citing rapport, where the record is not clear as to whether the trial court relied upon or rejected that reason, under AEDPA, we cannot say it was unreasonable for the state court to find no constitutional violation of Batson.
. The written question the prosecutor references in the fourth justification asks: “Do you feel a teenager is any more or less believable as a witness than an adult?” The written question the prosecutor references in the fifth justification asks: “Do you have any opinions that the victims of sexual assault are more or less believable than those who report being the victim of other crimes?”