dissenting.
I am in general agreement with Judge Peck’s analysis of the violation of the defendant’s privilege against self-incrimination. While the prosecutor’s remarks were not wholly gratuitous, a careful examination of the factors set forth in Hearn persuades me that the remarks did violate defendant’s rights, and that the prosecutor *978should not have engaged in this line of rhetoric. However, I disagree with the court’s harmless error analysis.
An examination of the issue of harmless error always presents a somewhat speculative undertaking. It is always conceivable that any factor may have affected the jury’s deliberation. However, over the years courts have developed a number of guideposts in undertaking a task which essentially involves looking into the minds of the jurors to examine the evidence with an eye to whether a decision might have been influenced by impermissible conduct. I agree with Judge Peck’s quotation of the standard from Chapman as to determining whether this is a case in which “absent the constitutionally forbidden comments, honest, fair-minded jurors might very well have brought in not-guilty verdicts.” 386 U.S. at 25-26, 87 S.Ct. at 828-29. For the reasons I will set forth below, I believe that there is no reasonable doubt that the same honest, fair-minded jurors who convicted Lent would not also have convicted him in the absence of the prosecutor’s impermissible arguments.
This was essentially a case of the victim’s word against the defendant’s, but with a number of corroborative physical elements of evidence. The prosecution had the direct testimony of the victim, and there was no question as to the identity of the other party to the events, only as to the exact nature of the events. If the jury believed the complainant, the case was essentially over.
The jury had ample opportunity to observe the victim’s demeanor and truthfulness. She was vigorously cross-examined on the obvious grounds of her conduct earlier in the evening, including going to a bar (albeit a campus bar), lying about her age to obtain alcohol, and leaving the bar, taking with her in her car a man she had just met.1 While these events may have been unwise, they constitute no defense to a charge of rape, if the victim’s version of the events is believed.
The victim’s version was particularly corroborated by the following physical items:
(1) there was testimony and medical evidence as to physical damage to her face, though defendant’s statement, as related by the police, was that he never injured her;
(2) a police officer indicated that, in his opinion, the condition of the grass around the car was more consistent with a struggle than with a consensual sexual encounter;
(3) her ear was ripped and bloody and her earring was missing;
(4) the victim’s purse was found under the matted grass outside the car, inconsistent with a consensual encounter; and
(5) the victim’s story emphasized that her conduct was motivated by a desire to maintain her virginity, and that virginity was confirmed by medical evidence.
Thus, the prosecution’s case was direct and well supported, if the jury found the victim credible. The jurors obviously did, as it would have been easy to return a not-guilty verdict had they had significant doubts as to her version of the events. In short, defendant’s decision not to take the stand, and the prosecution’s impermissible comments on it, would not seem to add much to the weight of the prosecution’s case.
I would agree that this is not a case in which the evidence could be termed “overwhelming,” as it might be in some cases with numerous neutral eyewitnesses and scientific identification testimony. At the same time, the harmless error analysis must also focus on the weightiness of the particular error. Comment on failure to take the stand might be more damaging in an alibi defense rather than where, as here, a straight-forward swearing contest is involved. Defendant’s version of the events was before the jury through the police officer’s statement. His decision not to display his own demeanor to the jury had to *979be extremely obvious to the jury independent of any prosecutorial comments.
It is also important to note that the defense attorney’s argument to the jury directly contradicted the defendant’s statement as relayed by the police officer. According to the police officer, Lent agreed that penetration had occurred, but that it had been wholly consensual and no injury had been inflicted. The defense counsel specifically said in the opening statement that there had been “petting, making out or whatever,” with no reference to sexual contact, and that after the victim bit the defendant’s tongue because of his persistent advances:
Greg, probably feeling the effects of the alcohol, as was Miss Thering, overreacted and struck, slapped Miss Thering one time across the right cheek. Following that a scuffle ensued between the two. That is probably when the earring — the injury to the ear. Following the scuffle, Greg got up, and he proceeded to walk along Lincoln Road towards his home.
In other words, the defense opening statement account was completely inconsistent with the contemporaneous account of the defendant as related by the police officer. This inconsistency also makes it much less likely that the jury was motivated by the prosecution’s improper remarks rather than by the strength of its case and the implausibility of the defense.
I also wish to comment on the point made in the majority’s opinion that the jury would have had to struggle with “doubts” that may have arisen due to petitioner’s seemingly inconsistent actions. These actions are presumably the matters referred to at page 977: that defendant had originally shown the victim his driver’s license, and that he was seen by others leaving the scene in a casual manner, without trying to conceal his identity.2 It is true that these pieces of evidence are more consistent with what has become known as “acquaintance rape” than with the more lurid image of a random attack by a stranger on the street or at home. However, those items of evidence are in no way inconsistent with the victim’s version of events, which the jury believed. Neither do they mitigate the ultimate seriousness of the crime.
The fact that there was some evidence which “might also be construed by reasonable jurors as evidence of petitioner’s innocence” (page 977) does not terminate the harmless error analysis. In every case, there is some evidence which if believed (favorable to the defense) or disbelieved (favorable to the prosecution) could lead to acquittal. The question is rather how those credibility judgments and juror judgments interact with the constitutional violation. In this case, I am firmly persuaded that the jury’s analysis of the facts which led to their guilty verdict did not depend on any additional discounting of defendant’s arguments (offered either through the police or through defense counsel) caused by the prosecutor’s “piling on” about the failure to testify.
Finally, I would note that an analysis of cases in which this particular constitutional violation is alleged reveals an almost complete absence of cases in which a conviction has actually been reversed. Out of some 50 cases I have examined in which a constitutional violation was considered, only two convictions were actually overturned. In Raper v. Mintzes, 706 F.2d 161 (6th Cir.1983), this court considered a murderous attack in the home on a recently separated woman, her new boyfriend and two of the boyfriend’s children. Only the boyfriend survived. By a 2-1 vote, the court upheld the district court’s grant of habeas corpus on the conviction for the first degree murder of the defendant’s wife, but affirmed the other convictions of second-degree murder of the children and assault with intent to murder the boyfriend. Id. at 167. The prosecutor’s comments as to the failure to explain what was in the defendant’s mind were particularly egregious because of the importance of mental state in first degree murder. Id. at 163.
*980In Hearn v. Mintzes, 708 F.2d 1072 (6th Cir.1983), the court reversed a conviction in another “acquaintance rape” situation. However, an examination of the facts in that case highlights the differences that would make the error here harmless beyond a reasonable doubt. In that case, the complainant’s conduct after the charged events gave reason to doubt her credibility. She told someone the night of the assault that she was “knocked down” with no mention of forced sex; the next day she told her employer that she was assaulted in an elevator; and when she first reported it to the police, she did not mention sexual contact. Id. at 1074. The defense also introduced five photographs of “an outwardly relaxed 'complainant' seated in the proximity of a door to the hallway, with a highball in her hand.” Ibid. The court emphasized that “the testimony of [complainant] was in many respects inconsistent; other witnesses strongly suggested that [complainant] knew [defendant] and willingly went to his apartment after embracing him in the hall.” Id. at 1078. Thus, the victim was far less credible than in the present case.
Rape under circumstances such as these can easily be disbelieved, and convictions are not easy to obtain. The victim’s degree of credibility is demonstrated by the fact that a conviction resulted. See Estrich, Rape, 95 Yale L.J. 1087, 1143 (1986). In our case, the substantial physical evidence verifying the victim’s story of a forceable assault and the contradictions in the defense version of events all point to an error that was harmless here, as opposed to the errors in the two cases above. I would affirm.
. Her girlfriend and defendant's friend went in defendant's car, all in search of a supposed party. The four young people intended to meet again at the party.
. In fact, he was walking on the side of a rural road at 4 A.M. on a cold night with no shirt on. The fact that he did not attempt to conceal himself from passing motorcyclists is hardly a powerful rebuttal of guilt.