dissenting.
The prosecution committed flagrant violations of Basel Jaradat’s constitutional right to remain silent, thereby completely obliterating his credibility. The majority chastises the prosecutor for his misconduct but allows the conviction to stand. Because the jury rejected much of the victim’s testimony, the prosecutor’s comments went directly to the contested issue; and because the physical evidence is not free from ambiguity, I respectfully dissent.
The majority opinion effectively holds that the presence of Jaradat’s semen on the vaginal swab combined with Jaradat’s denial of vaginal penetration means that any error in Jaradat’s trial must necessarily be harmless. The case is simply not so clear cut and is rife with ambiguities that leave me, at a minimum, in “grave doubt” about whether the blatant violations of Doyle v. Ohio, 426 U.S. 610, 96 S.Ct. 2240, 49 L.Ed.2d 91 (1976), had a “substantial and injurious” impact on the guilty verdict. Brecht v. Abrahamson, 507 U.S. 619, 628, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993). See also, O’Neal v. McAninch, 513 U.S. 432, 435, 115 S.Ct. 992, 130 L.Ed.2d 947 (1995) (finding that where a judge was in “grave doubt” about the harmlessness of *872the error, “the uncertain judge should treat the error, not as if it were harmless, but as if it affected the verdict”).
The majority, like all courts which have reviewed the record of the case, acknowledge that Doyle violations were committed. Respondent even agrees that a Doyle violation occurred. The sole question, therefore, is whether or not this clear constitutional error was merely “harmless.” As the majority notes, Jaradat was accused of five counts of rape: (1) fellatio, (2) cunnilingus, (3) anal penetration, (4) digital, vaginal penetration, and (5) vaginal intercourse. He was only convicted of vaginal intercourse. The majority finds the Doyle violation “harmless” because of the physical evidence that Jaradat’s semen was found in the victim’s rape kit.1
The crux of the matter is, somewhat tellingly, explained best by the prosecution. In its brief, Respondent writes: “One sensible reading of the jury’s verdict is this: In this case of classic he-said-she-said, the jury didn’t know whom to believe. So they believed the evidence.” (Respondent Br. at 26). Under the prosecution’s theory, the jury needed to reject the credibility of both witnesses. Since the prosecution committed a blatant Doyle violation designed to destroy Petitioner’s credibility, it seems preposterous to find that it is harmless error when even the Respondent admits that the jury’s verdict was based on rejecting the credibility of both witnesses. Crucially, neither the majority nor Respondent point to other evidence as to why Jaradat’s credibility should be questioned.
Several additional unique factors about the case further emphasize the way that the impermissible attack on Jaradat’s credibility caused “a substantial and injurious effect or influence in determining the jury’s verdict.” Kotteakos v. United States, 328 U.S. 750, 776, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946). Perhaps the most important is that the jury undoubtedly rejected much of the victim’s testimony. She alleged five different sex acts, and the jury found for Jaradat on four of the five. With respect to two of these acts, Jaradat had argued consent, and with respect to two he argued that the acts did not occur. Thus, the jury found for Jaradat both where he argued consent and where he argued the acts did not occur. The jury obviously rejected at least some of Jaradat’s testimony as well, since he denied the vaginal rape charge. The fact that the jury convicted Jaradat of the vaginal rape charge, however, does not in any way mean that it believed the victim’s version of events. The majority states that the presence of Jaradat’s semen in the victim’s vagina “confirms the testimony” of the victim. (Majority Op. at 870). Considering the wholesale rejection of every other part of the victim’s testimony, the presence of semen in the victim’s vagina does not in any way “confirm” her story.
The majority’s effort to link the physical evidence to the defendant’s testimony highlights how even with the physical evidence, the Doyle violations in this case were not harmless. The jury’s verdict is not easily condensed into one understanding of what happened. Lower courts attempted to explain the jury’s verdict with the incorrect statement that the jury found against Jaradat only where he did not argue consent. Respondent argues that the jury rejected all testimony and trusted the physical evidence. The majority appears to adopt this same reading of the *873jury’s verdict.2 The simple truth is that it is impossible with the mixed verdict in this case to understand what exactly motivated the jury. This same verdict could be reached by juries believing very different series of events. For instance, this verdict is consistent with a jury who may have thought that Jaradat committed all the charged offenses but were unable to find so “beyond a reasonable doubt” without corroborating physical evidence. At the same time, a jury could have flatly rejected the victim’s testimony but found unconvincing Jaradat’s hypothesis as to how his semen got into her vagina.
While divining what exactly a jury believed may be a fruitless task, the second outcome appears more likely. Petitioner asserts, with no rebuttal from Respondent, that the jury initially voted 11-1 to acquit on all charges and only convicted after three days of deliberation. In that situation, it seems plausible that at least one juror disbelieved Jaradat because of the Doyle violation. A juror could have found Jaradat more believable than the victim on the stand but reasoned that if Jaradat’s current explanation was true, he would have come forward earlier. The reason a prosecutor commits a Doyle violation is to induce a juror to make just this calculation. The prosecution was attempting to undermine the defendant’s credibility, and once Jaradat’s credibility was called into question, it became difficult for a juror to believe his explanation for the rape kit finding his semen on the vaginal swab.
This scenario is all the more problematic because, despite the majority’s assertions to the contrary, part of the Doyle violation dealt with the contested issue. The prosecutor in closing argument specifically referenced Jaradat’s exercise of his Miranda rights on the issue of how his semen got into the victim’s vagina.
He spoke so much English that on that day he neglected to tell anyone on December 28th oh, by the way, I performed oral sex on her, she performed oral sex on me, she took $40 from me I gave — -never mentioned anything of that on December 28th. But, see, on March 29th, when we have the DNA, your semen, oh let’s see what else is he going to say? It’s not mine? Of course it’s yours. Okay. I got another argument. She must have inserted it in her, and oh, I forgot to mention this thing about the $40 dollars and the sex in exchange.
(Trial Tr. 697-98). In no way is that colloquy, as the majority states, an emphasis on “the inconsistency between Jaradat’s testimony at trial that he never had sexual intercourse with the victim and the presence of his semen in her vagina.” (Majority Op. at 870). The first and last sentences of the paragraph are tied together with their reference to the oral sex and the $40. It strains credulity to suggest that the middle part, emphasizing Jaradat’s explanation for the semen in her vagina, is not also part of the Doyle violation. The entire sequence is specifically designed by the prosecutor to undermine Petitioner’s testimony on these issues be*874cause Petitioner had the opportunity to come forward with this information on December 28th.
The prosecutor in closing was undoubtedly making the point that Jaradat did not develop his theory for how the semen was found inside the victim until after he knew about the results of the rape kit. The prosecutor intentionally and impermissibly questioned Jaradat’s credibility on the issue based on his failure to come forward initially with this theory. No other reading of this passage from the closing argument is plausible; it was clearly intended to cast doubt on Jaradat’s explanation as to how his semen was found in the victim’s vagina by emphasizing that Jaradat did not come forward with this story immediately. The Doyle violation in the closing argument therefore directly related to the contested issue of the vaginal rape. Since Jaradat’s explanation may appear somewhat suspect, a Doyle violation in direct reference to that explanation seems particularly damaging and undoubtedly had a “substantial and injurious” influence on the jury’s decision.
The majority misapprehends this argument, believing that I think the jury “based” its decision on the prosecution improperly commenting on Jaradat’s silence even though it is likely that the jury “based” its decision on the totality of what was presented at trial. This includes both the physical evidence and the improper comments on Jaradat’s silence. The majority is correct that the difference between the vaginal rape and the other charges is the physical evidence, but the physical evidence alone was not, in and of itself, proof of guilt. Presumably, the jury also had to reject Jaradat’s testimony, something that was much easier to do after the prosecutor violated Jaradat’s right to remain silent. The majority appears to believe that had Jaradat been convicted of all counts, his habeas petition would be stronger. I disagree. The finding of not guilty on the other four counts does not mean that the jury was unmoved by the Doyle violations. Rather, the four not guilty verdicts more likely indicate that the jury found the victim not credible and increase the likelihood that the prosecution’s improper comments had a “substantial and injurious effect or influence in determining the jury’s verdict.” Kotteakos, 328 U.S. at 776, 66 S.Ct. 1239.
The inconsistency of the verdicts and the relationship of the Doyle violations to the conviction are potentially important enough on their own to show that the error was not harmless. Additionally, however, the physical evidence so central to the majority opinion’s conclusion is not without its own ambiguities. While the parties dispute whether Jaradat vaginally penetrated the victim, both agree that he did not ejaculate in her. Therefore, the mere presence of semen in her vagina is not conclusive proof of vaginal intercourse. The victim testified that Jaradat put his penis in her vagina “maybe three or four times.” (Trial Tr. at 468-69). Later, she testified that his penis was in her vagina “not long.” (Trial Tr. at 498). She testified that he then removed his penis and ejaculated on her “stomach,” which she later clarified as “my pubic area right underneath my stomach.” (Trial Tr. 468-69). Jaradat denied any vaginal intercourse, which means the two witnesses to the event both agreed that Jaradat did not ejaculate in the defendant, and even under the victim’s testimony, he had entered her for “not long.”3
*875Therefore, even if the jury accepted the victim’s version of the story, the presence of semen in her vagina was not the result of ejaculation. These facts were ignored by the state appellate court, which held that “Jaradat’s semen inside the victim’s vagina, and Jaradat’s denial of vaginal intercourse, but admission of other sexual conduct, constitutes overwhelming proof of Jaradat’s guilt.” The magistrate judge disagreed that the case was so clear but also relied on the DNA evidence without considering where Jaradat ejaculated, finding that the jury’s guilty verdict on vaginal intercourse meant that it did not rest “solely” on the issue of credibility. The magistrate judge stated only that the “existence of alternative arguments regarding the import of that physical evidence [does not] sufficiently undermine the likelihood that the jury was able to reach its verdict without being unduly influenced by the prosecution’s improper conduct.”
The district judge finally confronted the fact that the victim testified that Jaradat did not ejaculate inside her. The district judge held that “Petitioner was not wearing a condom and fails to consider the possibility that pre-ejaculate may have remained in the victim’s vagina after penetration, even if he ejaculated elsewhere.” Nobody disputes the “possibility” that preejaculate could account for the semen on the vaginal swab. On harmless error review, however, a court’s task is not the same as on a sufficiency of the evidence claim. “The inquiry cannot be merely whether there was enough to support the result, apart from the phase affected by the error,” but instead “whether the error itself had substantial influence.” Kotteakos, 328 U.S. at 765, 66 S.Ct. 1239.
Perhaps it is here where I most diverge from the majority opinion. The majority asserts that: “There is no other explanation, except rape, for how Jaradat’s semen was found inside the victim.” (Majority Op. at 870). Not even Respondent believes that the physical evidence is that strong. While questioning the reasonableness of Petitioner’s theory, Respondent states: “Is it possible that [the semen] ‘dripped’ into the victim’s vaginal canal after Jaradat ejaculated on the victim’s stomach as he claims? Sure.” (Respondent Br. at 24). Respondent argues that the jury, however, “was entitled to find that the semen found inside the victim’s vagina was proof that a vaginal rape with Jaradat’s penis occurred.” Id. While the jury was entitled to find that it was proof of vaginal rape, Jaradat was entitled to not have his post-arrest silence used against him, and he was further entitled to not have the impermissible use of his post-arrest silence influence the jury.
Here, even under the victim’s version of events, the physical evidence could not constitute a large amount of semen. While pre-ejaculate is a reasonable explanation, it seems possible that the “leakage” argument made by Petitioner also could have led to the positive rape kit result. Petitioner testified at trial that following fellatio, he ejaculated into his own hand and that it some of it got on the victim’s body. He then gave her napkins to clean up and hypothesizes that some of it got into her vagina. Contrary to the majority’s assertion, this theory is not inconsistent with the “leakage” argument emphasized by Petitioner’s counsel at oral argument. The cohesive theory is that Jaradat ejaculated on the victim, and in the process of cleaning up, semen got into her vagina. This may not be the only possible explanation, but it is bolstered by the fact that seminal fluid was also found on the victim’s rectal swab and a chair in the room. Respondent never suggested that Petitioner’s anal penetration occurred with his penis, so the presence of seminal fluid on the rectal exam supports Jaradat’s *876leakage theory. Jaradat equally denied anal penetration and was found not guilty of that offense.
With all of these irregularities, the majority is forced to base its entire holding on the vaginal swab. Nobody disputes that the vaginal swab is evidence that Jaradat penetrated her vaginally, but our inquiry is not whether sufficient evidence supported the jury’s verdict but whether the blatant and repeated Doyle violations had a “substantial and injurious” effect on the jury’s decision. Brecht, 507 U.S. at 623, 113 S.Ct. 1710. The rejection of much of the victim’s testimony, the presence of semen on the rectal swab, the fact that the Doyle violations explicitly invoked Jaradat’s explanation for his semen in the victim’s vagina, and the consensus that Jaradat did not ejaculate in the victim all combine to show that the prosecutorial misconduct had a “substantial and injurious effect or influence in determining the jury’s verdict.” Kotteakos, 328 U.S. at 776, 66 S.Ct. 1239.
For these reasons, I respectfully dissent.
. The majority also references the fact that Jaradat locked the door, turned off the lights, and closed the store. While these are actions that could be taken by a man planning a rape, they also seem like reasonable steps that might be taken by someone before engaging in consensual oral sex, which is what Jaradat testified occurred.
. The majority is only able to construct an argument that the Doyle violations were harmless, alone or in combination with the prosecution's overreaching attacks on Petitioner’s credibility, by engaging in a surfeit of speculation regarding the impact of the physical evidence, or the Petitioner's silence, on the jury's deliberations. The majority’s unsupported contention that the jury likely responded in a discernible and predictable way to the numerous Doyle violations and attacks on Petitioner’s credibility would be questionable under the best of circumstances. In the instant case, the majority's implied ability to invade the jury's consciousness, or the confines of the jury room, in order to divine the jury's thought processes and deliberations, seems ridiculous in the extreme.
. Obviously, "not long” would still be rape if it was without consent, but the relevant factor for this analysis is how Jaradat’s semen got into the victim’s vagina.