(concurring in part, dissenting in part).
While I concur with the majority’s decision that the evidence was sufficient to prove that Quick acted with premeditation and without heat of passion, and that Quick’s pro se arguments are without merit, I respectfully dissent on the determination that the district court properly excluded the testimony of three defense witnesses as being irrelevant, and that such exclusion was harmless error. I would conclude that the testimony of the three defense witnesses was relevant to the issue of premeditation and that the exclusion of their testimony denied Quick his constitutional right to present a defense and was prejudicial error.
I.
The majority opinion recognizes that “Quick’s state of mind during the summer [of 2000] could be relevant in determining whether Quick premeditated the murder of Mueller,” but ultimately concludes that the district court did not abuse its discretion in excluding evidence on Quick’s state of mind as irrelevant. The majority mentions four grounds to support the district court’s discretion: (1) the testimony con*721cerned Quick’s state of mind “months before the shooting”; (2) the testimony concerned “the relationship between Quick and his wife,” not the relationship between Quick and Mueller; (3) the testimony was cumulative to “the testimony of Quick, his wife and his wife’s brother”; and (4) the transcripts of the police interviews with the three witnesses, which constituted Quick’s offer of proof, did not support Quick’s claim that they would show his lack of anger. Before addressing each of these four grounds, it is important to consider precisely what grounds were relied upon by the district court, whose discretion is being reviewed.
The state’s original pretrial motion in limine to exclude the testimony of these three witnesses was not based solely, or even primarily, on grounds of irrelevance, but on grounds that the three witnesses might offer character evidence- concerning Diane Quick and Justin Mueller. In fact, the district court characterized the state’s motion as one “to exclude evidence of the character of Diane Quick or Justin Mueller.” The state did argue that the transcripts of interviews with these three witnesses concerned events that “were all somewhat distant in time from the homicide,” but it also expressed concern with the statements about the type of person Diane was, which was inadmissible as character evidence. Quick’s counsel explained that the evidence would focus on Quick’s state of mind: “[w]e do not intend * * * to attack Mr. Mueller or Ms. Quick,” but instead “[w]e’re going to talk about Jon, the way Jon was handling this, and his reluctant acceptance of the divorce, his reluctant acceptance of the circumstances, his reluctant acceptance that he was going to have less contact with his children. And that will lay the foundation for Mr. Quick’s' ultimate testimony in this case.”
The district court did not rule on that motion in limine prior to trial, but waited until the state’s case was almost complete. By that time, Diane Quick had already testified and the state acknowledged that any' concerns about character evidence had become moot. The state also agreed that it would be fair to characterize the proffered testimony as bearing on “defendant’s state of mind at the time he spoke with these individuals.” The court then announced its decision, only stating “I’m not going to allow that evidence because I don’t think it’s relevant.” No further explanation was given and no specific grounds to support the relevancy determination were mentioned. Thus, we can only speculate about how the court exercised its discretion or what factors informed that discretion. And, it must be underscored, the district court decided to exclude this testimony before it heard the testimony of Quick and before it was certain that Quick would testify in his own defense.
A. Months before the Shooting
Perhaps the district court considered, as the majority opinion speculates, that the proffered testimony was irrelevant because it concerned Quick’s state of mind at least one month before the shooting and Quick’s state of mind “could have changed substantially” during this month. But this consideration goes to the weight of the proffered testimony, not to its admissibility.
First, the relevant timeframe is quite short to begin with. ■ The shooting occurred on September 14, 2000, and the parties had only separated in May 2000, for a total timeframe spanning only about four months. The interview transcripts describe conversations that occurred with Quick’s first marriage counselor in three meetings on June 23 and 30 and July 7; with his second -marriage counselor in *722three meetings on July 13 and 18 and August 1; and with his attorney in at least three meetings on July 13, August 15 and September 6. Thus, the first marriage counselor had the opportunity to observe the evolution of Quick’s mental and emotional state to within two months of the shooting; the second marriage counselor had the opportunity to observe the evolution of Quick’s mental and emotional state to within 45 days of the shooting; and the attorney had the opportunity to observe the evolution of Quick’s mental and emotional state to within eight days of the shooting.
In a first-degree murder trial, where the defense admits that the defendant was the shooter and premeditation is the critical issue, it seems highly unprodhctive to apply a heightened relevancy standard to screen the defendant’s testimony on mental and emotional state simply because it is not tip to the minute of the shooting. Viewed from another perspective, while the state offered anecdotal evidence that indirectly bore on Quick’s state of mind, the court excluded the testimony of disinterested professional witnesses who would have directly- addressed Quick’s state of mind.'- And Quick’s last meeting with his attorney occurred after the events described in the state’s evidence.
B. Relationship with Diane Quick,-not Mueller.
If evidence of a relationship between a defendant and the victim is generally admissible to show premeditation, then surely. evidence about the relationship between a defendant and his estranged wife, who had become romantically involved with the victim, is equally relevant. It is unrealistic to separate Mueller from Diane Quick in this analysis. Quick shot Mueller when he found him in an intimate setting with his wife. Of course, Quick’s issues were with his wife, not directly with Mueller, but it is not a leap in logic to conclude that his jealousy over his wife’s conduct would be directed against the stranger who had taken his place in his wife’s life.
More importantly, the proffered testimony was not focused on the relationship between defendant and Diane Quick (particularly after Quick’s counsel made it clear that he was not going to attack the character of Diane Quick), but on Quick’s state of mind as a result of the break up of the relationship.
C. Cumulative to Testimony of Quick, Diane Quick and Fredrick Flood.
The district court could not have considered the proffered testimony of these three witnesses to be cumulative of the testimony of Quick because it ruled to exclude that testimony before the state rested. Quick had not yet testified and he retained the right under the Fifth Amendment to decline to testify. Further, a criminal defendant is denied the constitutional right to defend himself if he is restricted to his own testimony and not permitted to offer corroborating evidence. Obviously, the exclusion of third-party testimony to explain a defendant’s state of mind puts undue pressure on the Fifth Amendment rights of the defendant.
As to Diane, she was a hostile witness to Quick. She testified that Mueller was the best friend of her brother, Fredrick Flood; that their relationship built incredibly fast after her separation from Quick; and that they intended to be married. She then witnessed Quick shoot Mueller in her own home, while her three children were sleeping in nearby rooms.
Further, Diane had obtained an order for protection against Quick in June of 2000 and did not testify to any face-to-face meetings or even live telephone conversations .with Quick from that date to the date *723of the shooting. Her communications were only by note, email or voice message. Thus, during the entire time that Quick was meeting with the three proffered witnesses, and discussing his mental and emotional state, Diane Quick had no direct contact with him.. Obviously, she did not corroborate Quick’s testimony.
As to Diane’s brother, he likewise was hostile to Quick. He acknowledged that Justin Mueller had been a co-worker and was his best friend. He said that he was excited that Mueller and his sister, Diane, were dating. He testified to only one communication with Quick in the relevant time period and the state offered that-testimony to show premeditation.
Neither Diane Quick nor Diane’s brother testified directly about their observations of Quick’s mental or emotional state in the summer of 2000.
D. The Proffered Testimony was not Supportive of Quick’s Defense.
As a matter of convenience, Quick’s counsel used the transcript of interviews with these three witnesses as the offer of proof. While these transcripts adequately supply the substance of what each witness would testify, they obviously do not reflect the precise way that Quick’s counsel would want to develop that testimony. That being said, the transcripts do describe, a progression in Quick’s mental and emotional state, from denial, to self-blame to acceptance. The general tone of the interview transcripts is that Quick was approaching this issue calmly, without signs of anger. To the extent there are statements that might contradict that tone, or tend to show anger, those statements do not render the entire testimony irrelevant but simply provide fruit for cross-examination.
In sum, none of these grounds supports the exclusion of this testimony.
II.
The majority opinion goes on to conclude that, even if erroneous, the exclusion of the testimony would be “harmless beyond a reasonable doubt.” In my view, the harmless error analysis should only rarely be employed in a situation where a defendant has been denied his constitutional right to present a complete defense. Further, the use of a harmless error analysis to resolve criminal appeals often allows the prosecution to have it both ways. First, the prosecution fights to exclude evidence favorable to a defendant and then, when successful, argues on appeal that the evidence would not have been that important and surely would not have changed .the verdict. Obviously, the state perceived the testimony of the three defense witnesses to be prejudicial to the state, so much so that it pursued a motion in limine before the witnesses could even be called to the stand. Simultaneously, the state planned to offer its own evidence on premeditation, leaving Quick with the only option of supplying his own testimony on the subject. I would take the state at their first word — the state’s motion in li-mine tells the court that the evidence would be prejudicial and the state’s argument on appeal should not be otherwise.
. Finally, I am not comfortable placing myself in the shoes of this jury and determining whether or not this relevant evidence could have affected the jury’s thinking on the issue of premeditation. Perhaps the state’s evidence was so overwhelming that a juror would not have been influenced. But jurors are not that predictable. Fortunately, jurors are not like computers that answer the same question in the same way every time. Instead, jurors exercise broad discretion in evaluating the evidence .to arrive at their ultimate conclusions.-
*724Accordingly, I would reverse the conviction and order a new trial to include the testimony of the three excluded defense witnesses.